Skip to main content

Foundational Trials

INTREPID

Image: Shutterstock / ChaNaWiT

Publication

  • Title: Fever Prevention in Patients With Acute Vascular Brain Injury: The INTREPID Randomized Clinical Trial
  • Acronym: INTREPID
  • Year: 2024
  • Journal published in: JAMA
  • Citation: Greer DM, Helbok R, Badjatia N, Ko SB, McKenna Guanci M, Sheth KN; for the INTREPID Study Group. Fever Prevention in Patients With Acute Vascular Brain Injury: The INTREPID Randomized Clinical Trial. JAMA. 2024;332(18):1525-1534.

Context & Rationale

  • Background
    • Fever and higher body temperature after acute vascular brain injury are consistently associated with worse outcomes (mortality and disability) across diverse neurological syndromes in observational meta-analytic evidence.1
    • Clinical practice commonly treats fever after it develops, yet the causal role of fever versus “severity marker” remains uncertain, and high-quality evidence for prophylactic fever prevention in vascular brain injury has been limited.
    • Temperature management (antipyretics, surface/intravascular devices) carries burdens (shivering, sedative/paralytic exposure, nursing workload, skin injury), making the risk–benefit balance clinically important.
  • Research Question/Hypothesis
    • Whether proactive, device-based fever prevention (targeted normothermia at 37°C) reduces fever exposure (“fever burden”) and improves patient-centred neurological outcomes compared with standard (reactive) fever management in ICU patients with acute vascular brain injury.
    • The trial protocol (including hierarchical outcomes and adaptive features) was published before completion.3
  • Why This Matters
    • If fever prevention improved disability or survival, it would justify routine, resource-intensive temperature management in neurocritical care.
    • If fever prevention reduced fever but did not improve outcomes, it would challenge the assumption that prophylactic normothermia is a high-value intervention and refocus efforts on targeted indications and minimising iatrogenesis.
    • Fever burden is a plausible biological mediator; INTREPID tests whether modifying this intermediate phenotype changes clinical endpoints.

Design & Methods

  • Research Question: In ICU patients with acute vascular brain injury and high risk of fever, does proactive fever prevention using a surface temperature management device (target 37°C) reduce fever burden and improve 90-day functional outcome compared with standard care fever management?
  • Study Type: Multicentre, international, randomised (1:1), open-label temperature management strategy trial with blinded assessment of functional outcomes; prespecified hierarchical testing and an adaptive sample size/futility framework; ICU (neurocritical care) setting.
  • Population:
    • Setting: ICU patients with acute vascular brain injury (ischaemic stroke, spontaneous intracerebral haemorrhage, aneurysmal subarachnoid haemorrhage).
    • Inclusion (core features): adults admitted to ICU with moderate–severe injury (ischaemic stroke NIHSS ≥6; intracerebral haemorrhage ICH score ≥2; aneurysmal subarachnoid haemorrhage WFNS grade ≥3); expected ICU stay ≥72 hours; core temperature <38.0°C at enrolment/confirmatory check; diagnosis-specific stabilisation windows (e.g., ischaemic stroke 3–24 hours from ictus; intracerebral haemorrhage 6–36 hours from last known normal; aneurysmal subarachnoid haemorrhage 4–48 hours after post-procedural stabilisation and within 72 hours of symptom onset).
    • Key exclusions (examples): fever ≥38.0°C at enrolment/confirmatory check or prolonged fever before enrolment; substantial pre-morbid disability (premorbid mRS ≥3 if age 18–80 years; premorbid mRS ≥1 if age 81–90 years); limitations of care expected to preclude protocol delivery; contraindications to device application or core temperature monitoring (Not reported in detail in the primary publication).
  • Intervention:
    • Fever prevention strategy: Arctic Sun 5000 surface temperature management system initiated promptly after randomisation, programmed to maintain 37°C continuously through day 14 or ICU discharge (whichever occurred first).
    • Core temperature monitoring (e.g., temperature-sensing urinary catheter or oesophageal probe) used for feedback control; protocolised shivering prevention/management including counter-warming and pharmacologic shiver control as needed.
  • Comparison:
    • Standard care fever management: no prophylactic device-based temperature control; fever treated when it occurred using a prespecified tiered algorithm (pharmacologic antipyretics and surface cooling first, escalating to advanced temperature management devices when needed).
    • Rescue/adjuncts (including targeted temperature management devices) permitted according to the algorithm when clinically indicated.
  • Blinding: Treating teams and patients were not blinded to strategy; 90-day functional outcomes were assessed by certified assessors blinded to allocation; data analysts/statisticians were blinded during analysis (as reported).
  • Statistics: Power calculation: 1,000 patients were required to detect a 10% absolute increase in favourable 90-day outcome (mRS 0–3) from 40% (standard care) to 50% (fever prevention) with 88% power at a 2-sided 5% significance level; inflating for ~15% attrition yielded a planned sample size of 1,176; prespecified interim assessment at 588 treated patients with potential sample size increase up to 2,000 evaluable depending on conditional power. Primary analysis was intention-to-treat with hierarchical testing (primary fever burden then principal 90-day mRS outcome); effect estimates and confidence intervals as reported.
  • Follow-Up Period: Acute physiological endpoints through day 14 (or ICU discharge), with clinical follow-up at 3 months (primary clinical time-point) and additional follow-up at 6 and 12 months (Not reported in detail in the primary publication results tables).

Key Results

This trial was stopped early. Stopped after the planned interim analysis (at 588 treated patients completing the 3-month assessment or discontinuing) for futility of the principal 90-day functional endpoint (conditional power 1.9%); 677/1,176 patients were enrolled when the trial stopped.

Outcome Fever prevention Standard care Effect p value / 95% CI Notes
Daily mean fever burden (°C-hours >37.9°C) through day 14 or ICU discharge (primary outcome) 0.37 (SD 1.0); n=320 0.73 (SD 1.1); n=329 Difference −0.35 95% CI −0.51 to −0.20; P<.001 Objective physiological endpoint; analysed with prespecified methods; fewer than 677 contributed due to missing temperature data/time at risk.
Total fever duration through day 14 or ICU discharge 9.0 hours (SD 19.41); n=320 21.6 hours (SD 29.81); n=329 Difference −12.4 hours 95% CI −16.2 to −8.7; P=Not reported Demonstrates reduced time febrile; derived from prespecified temperature definitions.
Any fever burden (>0 °C-hours) 175/320 (54.7%) 248/329 (75.4%) RR 0.73 95% CI 0.65 to 0.82; P<.0001 Binary indicator of any fever exposure; supports biological separation.
Modified Rankin Scale (mRS) at 90 days (principal secondary; shift analysis) Median 4.0 (IQR 3–6); n=278 Median 4.0 (IQR 2–6); n=306 OR 1.09 95% CI 0.81 to 1.46; P=.54 mRS 5 and 6 combined for modelling; blinded assessors; missing mRS data differed between groups.
Favourable functional outcome at 90 days (mRS 0–3) 109/278 (39.2%) 131/306 (42.8%) RR 0.92 95% CI 0.75 to 1.12; P=.40 Direction favoured standard care; confidence interval compatible with modest benefit or harm.
Death through 3 months 75/339 (22.1%) 73/338 (21.6%) Not reported Not reported Mortality similar; trial not powered for mortality differences after early stopping.
Any shivering (acute phase; as-treated) 283/331 (85.5%) 82/337 (24.3%) RR 3.52 95% CI 2.89 to 4.28; P<.0001 Substantial treatment burden; counter-warming and pharmacologic shiver control commonly used.
Significant shivering (BSAS ≥2; acute phase; as-treated) 90/331 (27.2%) 17/337 (5.0%) RR 5.41 95% CI 3.32 to 8.81; P<.0001 Clinically meaningful increase in shivering intensity with prophylactic device strategy.
  • Fever prevention achieved a clear reduction in fever exposure (daily mean fever burden 0.37 vs 0.73 °C-hours; total fever duration 9.0 vs 21.6 hours), but did not improve 90-day global disability (mRS shift OR 1.09; 95% CI 0.81 to 1.46; P=.54).
  • Subgroup fever-burden reductions were seen across syndromes: ischaemic stroke difference −0.10 (95% CI −0.35 to 0.15); intracerebral haemorrhage −0.50 (−0.78 to −0.22); subarachnoid haemorrhage −0.52 (−0.81 to −0.23) (all P<.001 as reported).
  • Intervention-related burden was large: any shivering 85.5% vs 24.3% and significant shivering 27.2% vs 5.0% (both P<.0001); device therapy was suspended at least once in 164/339 (48.4%) fever prevention patients with mean total suspension 628.6 minutes (SD 995.78).

Internal Validity

  • Randomisation and allocation:
    • Central randomisation with minimisation and stratification (site, diagnosis, age group, and baseline severity strata as reported).
    • Allocation concealment prior to assignment is supported by central randomisation (detailed mechanism beyond this not reported in the primary publication).
  • Drop out or exclusions:
    • Randomised: 339 fever prevention vs 338 standard care (ITT population).
    • Not treated: 9/339 (2.7%) fever prevention vs 1/338 (0.3%) standard care (discontinued before acute phase).
    • Premature discontinuation during study: 53/339 (15.6%) fever prevention vs 43/338 (12.7%) standard care.
    • Key 90-day mRS analysis included 278/339 (82.0%) fever prevention vs 306/338 (90.5%) standard care, indicating differential missingness for the principal clinical outcome.
  • Performance/Detection Bias:
    • Treatment teams were unblinded, creating potential for co-intervention differences (sedation intensity, ventilation, infection workup, use of cooling adjuncts).
    • Primary endpoint (temperature-derived fever burden) is objective and continuously measured, reducing detection bias for the main physiological endpoint.
    • Functional outcomes were assessed by blinded, certified assessors, mitigating detection bias for disability endpoints.
  • Protocol Adherence:
    • Fever prevention device delivered to most assigned patients (330/339 treated).
    • Device programmed to target 37°C in 306/339 (90.3%) fever prevention patients; counter-warming initiated concurrently in 259/339 (76.4%).
    • Therapy interruptions were common: suspension ≥1 episode in 164/339 (48.4%), mean total suspension 628.6 minutes (SD 995.78), consistent with real-world delivery constraints (procedures, shivering, skin tolerance).
  • Baseline Characteristics:
    • Diagnostic mix was balanced (intracerebral haemorrhage ~33%, ischaemic stroke ~38%, subarachnoid haemorrhage ~30%).
    • Temperature at randomisation was similar (mean 36.7°C in both groups as reported).
    • Severity strata were similar (high severity 46.2% vs 46.7%).
  • Heterogeneity:
    • Three syndromes with different fever mechanisms and outcome trajectories were combined; this improves scope but complicates inference about syndrome-specific benefit.
    • Fever burden reduction was present across syndromes, suggesting physiological effect generalised; clinical effect heterogeneity remains plausible but was not demonstrated by prespecified subgroup results.
  • Timing:
    • Enrolment required absence of fever at baseline and diagnosis-specific stabilisation windows; this improves protocol purity for “prevention” but may exclude the earliest highest-fever-risk period in some patients.
  • Dose:
    • The “dose” was targeted normothermia (37°C) continuously to day 14 or ICU discharge; achieved separation in fever burden and fever duration indicates the intervention was delivered at biologically meaningful intensity.
  • Separation of the Variable of Interest:
    • Primary separation: daily mean fever burden 0.37 (SD 1.0) vs 0.73 (SD 1.1); difference −0.35 (95% CI −0.51 to −0.20).
    • Exposure separation: total fever duration 9.0 (SD 19.41) vs 21.6 (SD 29.81) hours; difference −12.4 (95% CI −16.2 to −8.7).
    • Physiological cost of separation: any shivering 283/331 (85.5%) vs 82/337 (24.3%); significant shivering (BSAS ≥2) 90/331 (27.2%) vs 17/337 (5.0%).
  • Key Delivery Aspects:
    • Standard care group often received fever therapies (fever control therapy in 234/337 [69.4%]); advanced temperature management device use occurred in 34/234 (14.5%) of those receiving fever control therapy, potentially reducing between-group contrast for device exposure.
  • Outcome Assessment:
    • Primary outcome was algorithmic/temperature-derived and prespecified; principal clinical outcome used standard mRS with blinded assessment.
    • Major adverse events (death, pneumonia, sepsis, malignant cerebral oedema) were adjudicated by an independent clinical events committee (as reported).
  • Statistical Rigor:
    • Prespecified hierarchical testing and interim futility framework were applied; early stopping limits precision for clinical outcomes and increases susceptibility to chance imbalances in missing follow-up.
  • Funding / Sponsor Role (risk-of-bias consideration):
    • Industry sponsor provided funding/devices and operational oversight and statistical analysis; sponsor had a role in design, conduct, data collection, management, analysis, and interpretation; sponsor had no role in manuscript preparation/review/approval and decision to submit (as reported).

Conclusion on Internal Validity: Overall, internal validity is moderate: randomisation and objective temperature endpoints support causal inference for fever burden reduction, but open-label care, substantial shivering-related co-interventions, rescue temperature management in controls, and differential missing 90-day mRS data constrain confidence in clinical endpoint neutrality.

External Validity

  • Population Representativeness:
    • Represents a neurocritical care cohort: ICU patients with moderate–severe ischaemic stroke, intracerebral haemorrhage, or aneurysmal subarachnoid haemorrhage.
    • Excludes patients with fever at baseline and those with substantial pre-morbid disability, limiting applicability to frailer populations and those presenting already febrile.
  • Applicability:
    • Intervention requires continuous core temperature monitoring, surface temperature management devices, and structured shivering management (staffing and pharmacologic resources), most applicable to high-resource ICUs.
    • Standard care included a tiered fever algorithm; external validity depends on how closely other units mirror this structured approach.

Conclusion on External Validity: Generalisability is moderate to similar high-acuity neurocritical care settings with access to targeted temperature management and shivering-control protocols, but limited for non-ICU stroke populations, resource-limited settings, and patients presenting already febrile or with high pre-morbid disability.

Strengths & Limitations

  • Strengths:
    • International, multicentre randomised design across three major vascular brain injury syndromes.
    • Objective, continuously measured physiological primary endpoint with prespecified fever burden definition.
    • Blinded, certified assessment for functional outcomes; independent adjudication for major adverse events.
    • Pre-publication of the study protocol and planned analyses supports transparency and reduces analytic flexibility.3
  • Limitations:
    • Stopped early for futility, reducing precision for clinical endpoints and increasing vulnerability to missing-data mechanisms.
    • Open-label strategy with substantial co-intervention exposure (shivering management, sedation, procedures) that could offset or mask neurological benefit.
    • Differential missingness for 90-day mRS (278 vs 306 analysed), potentially biasing the principal clinical endpoint.
    • Control group access to advanced temperature management (including device use in a subset) may attenuate contrasts in “temperature management intensity”.

Interpretation & Why It Matters

  • Clinical practice
    • Routine prophylactic device-based fever prevention (target 37°C for up to 14 days) reduced fever exposure but did not improve 90-day disability or mortality, and substantially increased shivering-related burden.
    • These data support prioritising prompt detection and treatment of fever while weighing the harms and resource costs of aggressive prophylaxis.
  • Mechanistic inference
    • INTREPID challenges the assumption that reducing fever burden alone is sufficient to improve long-term neurological outcomes in heterogeneous vascular brain injury ICU populations.
    • The marked increase in shivering (and attendant pharmacologic management) highlights the importance of net clinical benefit rather than intermediate physiological targets.
  • Future trials
    • Findings emphasise the need for syndrome- and risk-enriched designs (e.g., patients with higher fever propensity or with specific injury phenotypes) and for strategies that minimise shivering and sedation confounding.

Controversies & Subsequent Evidence

  • Feasibility vs value:
    • Contemporary editorial appraisal emphasised that the trial demonstrates feasibility of preventing fever and reducing fever burden, but highlighted that clinical benefit was not shown and that shivering/management burden is substantial, complicating implementation at scale.4
  • Population selection and event rate considerations:
    • Correspondence questioned whether enrolment criteria (including anticipated ICU stay thresholds and exclusion of baseline fever) may have selected a population with lower modifiable fever-related risk, limiting potential for functional benefit.5
    • The trialists’ reply reiterated the rationale and design choices and emphasised that fever reduction was achieved without a signal of improved disability outcomes, reinforcing equipoise regarding prophylactic strategies.6
  • Interpretation of “negative” neurocritical care trials with strong physiological separation:
    • INTREPID adds to a pattern where modifying intermediate physiological variables (temperature exposure) does not necessarily translate into improved patient-centred outcomes once co-interventions and iatrogenic burdens are accounted for.

Summary

  • In 677 ICU patients with acute ischaemic stroke, intracerebral haemorrhage, or aneurysmal subarachnoid haemorrhage, prophylactic device-based fever prevention reduced fever burden and fever duration versus standard care.
  • The trial stopped early for futility on the prespecified 90-day functional endpoint; mRS shift did not differ (OR 1.09; 95% CI 0.81 to 1.46; P=.54).
  • Mortality through 3 months was similar (22.1% vs 21.6%).
  • Shivering was markedly increased with prophylactic fever prevention (any shivering 85.5% vs 24.3%; significant shivering 27.2% vs 5.0%).
  • Routine prophylactic device-based normothermia in unselected neurocritical vascular brain injury patients is not supported by these data; fever treatment strategies should consider net benefit and iatrogenic burden.

Further Reading

Other Trials

Systematic Review & Meta Analysis

Observational Studies

Guidelines

Notes

  • Across available stroke and haemorrhage guidelines, fever treatment is commonly recommended; however, post-INTREPID guideline statements specifically addressing routine prophylactic device-based normothermia for acute vascular brain injury were not identified within the provided trial materials.

Overall Takeaway

INTREPID demonstrated that prophylactic, device-based fever prevention in neurocritical care can reduce fever exposure, but this physiological success did not translate into improved 90-day disability or mortality and came with substantial shivering-related burden. The trial therefore reframes fever prevention as a strategy requiring careful patient selection and a clear net-benefit justification rather than a default ICU standard for all acute vascular brain injury patients.

Overall Summary

  • Prophylactic surface temperature management reduced fever burden (0.37 vs 0.73 °C-hours/day) and fever duration (9.0 vs 21.6 hours) but did not improve 90-day disability.
  • Trial stopped early for futility on functional outcome; effect estimates remained compatible with modest benefit or harm.
  • Shivering increased markedly (85.5% vs 24.3%), highlighting iatrogenic and implementation costs of aggressive prophylaxis.

Bibliography