Skip to main content

Foundational Trials

BOX - Blood Pressure

Image: Shutterstock / ChaNaWiT

Publication

  • Title: Blood-Pressure Targets in Comatose Survivors of Cardiac Arrest
  • Acronym: BOX (Blood Pressure Domain)
  • Year: 2022
  • Journal published in: New England Journal of Medicine
  • Citation: Kjaergaard J, Møller JE, Schmidt H, et al. Blood-pressure targets in comatose survivors of cardiac arrest. N Engl J Med. 2022;387:1456-66.

Context & Rationale

  • Background
    • Post–cardiac arrest syndrome commonly features myocardial stunning and vasoplegia, with hypotension early after return of spontaneous circulation (ROSC).
    • Observational datasets suggested an association between early hypotension (and/or lower mean arterial pressure, MAP) and worse neurological outcome, but causal inference is limited by confounding by illness severity and co-interventions.
    • Cerebral autoregulation may be impaired and/or right-shifted (particularly in chronic hypertension), creating physiological equipoise between “permissive lower MAP” and “vasopressor-driven higher MAP”.
    • Before BOX, randomised evidence for specific MAP targets after out-of-hospital cardiac arrest (OHCA) was sparse, small, and frequently used surrogate endpoints (e.g., biomarkers) rather than patient-centred functional outcomes.
  • Research Question/Hypothesis
    • In comatose adult OHCA survivors of presumed cardiac cause, does targeting a higher MAP (77 mmHg) compared with a lower MAP (63 mmHg) reduce death or severe neurological disability?
    • Hypothesis tested: a higher MAP target would improve survival with favourable neurological outcome by improving end-organ (especially cerebral) perfusion.
  • Why This Matters
    • MAP targets are a daily, titratable decision in post-arrest critical care and are strongly intertwined with vasopressor exposure, arrhythmia risk, afterload, and resource use.
    • Defining “how high is high enough” (or whether higher is beneficial at all) is central to standardising post-arrest bundles and to the design of future personalised perfusion trials.
    • Because post-arrest management is bundle-based, a blinded haemodynamic target strategy in a factorial platform provides a rigorous test of a single physiological lever within complex care.

Design & Methods

  • Research Question: Among comatose survivors of presumed-cardiac OHCA, does a higher MAP target (77 mmHg) compared with a lower MAP target (63 mmHg) reduce the risk of death or severe disability?
  • Study Type: Randomised, investigator-initiated, two-centre, parallel-group blood-pressure trial embedded within a 2×2 factorial design; blood-pressure intervention double-blind; conducted in specialised cardiac arrest centres (Denmark).
  • Population:
    • Adults (≥18 years) with sustained ROSC after OHCA, unconscious (Glasgow Coma Scale 3–8), presumed cardiac cause, and eligible for targeted temperature management; randomised within 4 hours after ROSC 1
    • Key exclusions included: unwitnessed asystole; suspected intracranial haemorrhage/stroke/intoxication; pregnancy; severe COPD; known limitation of care; MAP <50 mmHg; >4 hours from ROSC to randomisation 1
  • Intervention:
    • High blood-pressure target: actual MAP target 77 mmHg, achieved using a modified pressure module calibrated so displayed pressures were ~10% lower than measured (clinical teams targeted a displayed MAP of ~70 mmHg, yielding a higher actual MAP).
    • Vasopressor, inotrope, and fluid therapy were titrated by the treating team to meet the displayed target; the intervention ran while invasive blood pressure monitoring was used during ICU care.
  • Comparison:
    • Low blood-pressure target: actual MAP target 63 mmHg, using the same modified module calibrated so displayed pressures were ~10% higher than measured (displayed ~70 mmHg corresponding to a lower actual MAP).
    • Otherwise the same post-arrest ICU pathways and co-interventions were available (including temperature management, early coronary angiography/PCI where indicated, and standard organ support).
  • Blinding: Double-blind for blood-pressure assignment (clinical staff, patients, investigators, and outcome assessors masked via calibrated module); oxygenation assignment within the factorial platform was not blinded.
  • Statistics: Power calculation: 846 patients required to detect a 10% absolute reduction in mortality (from 38% to 28%) with 90% power at a two-sided 5% significance level; trial planned 800 participants. Primary analysis: Cox proportional-hazards model (site-stratified; adjusted for oxygenation assignment), intention-to-treat.
  • Follow-Up Period: Primary outcome assessed to hospital discharge within 90 days; functional and cognitive outcomes assessed at 3 months (90 days).

Key Results

This trial was not stopped early. The blood-pressure comparison completed enrolment and follow-up as planned (802 randomised; 789 included in the intention-to-treat analyses).

Outcome High BP target (MAP 77) Low BP target (MAP 63) Effect p value / 95% CI Notes
Primary composite: death from any cause or CPC 3–4 at discharge within 90 days 133/393 (34%) 127/396 (32%) HR 1.08 95% CI 0.84 to 1.37; P=0.56 Composite dominated by death (see next row); no interaction with oxygen assignment (P=0.89)
Death from any cause within 90 days 122/393 (31%) 114/396 (29%) HR 1.13 95% CI 0.87 to 1.47 Vital status obtained via registry and/or hospital records
Poor neurological outcome at 3 months (CPC 3–4) 31/393 (8%) 32/396 (8%) RR 0.98 95% CI 0.62 to 1.56 Telephone interview where feasible; some chart-derived assessments
Severe disability or death at 3 months (mRS 4–6) 134/393 (34%) 125/396 (32%) RR 1.09 95% CI 0.90 to 1.31 No difference in ordinal distributions of CPC or mRS reported
Renal replacement therapy initiated during ICU stay 41/393 (10%) 40/396 (10%) HR 1.03 95% CI 0.67 to 1.57 Event rate similar between groups
Mean arterial pressure during ICU stay (median, IQR), mmHg 76 (74–78) 72 (70–74) Not reported Not reported Achieved separation was larger early: mean difference 10.5 mmHg from 2–48 h
Norepinephrine total dose during ICU stay (median, IQR), mg 10.7 (3.5–24.2) 6.8 (1.8–15.0) Not reported Not reported Mean norepinephrine infusion difference (2–48 h): 0.038 µg/kg/min (95% CI 0.026 to 0.049)
Serious adverse events: infection 102/393 (26%) 110/396 (28%) RR 0.93 95% CI 0.75 to 1.16; P=0.45 No signal of increased serious infection with higher MAP
Serious adverse events: arrhythmia 59/393 (15%) 50/396 (13%) RR 1.19 95% CI 0.85 to 1.67; P=0.27 Despite greater vasoactive exposure in the high-target group, no statistically significant increase
Serious adverse events: any bleeding 82/393 (21%) 92/396 (23%) RR 0.89 95% CI 0.70 to 1.13; P=0.29 Uncontrolled bleeding: 22 (6%) vs 16 (4%); RR 1.39 (95% CI 0.76 to 2.55)
  • Physiological separation was achieved: mean MAP difference 10.5 mmHg (95% CI 9.9 to 11.2) between groups from 2–48 hours, at the expense of more vasoactive support (mean norepinephrine infusion difference 0.038 µg/kg/min; 95% CI 0.026 to 0.049).
  • No improvement was observed in the primary composite outcome (HR 1.08; 95% CI 0.84 to 1.37; P=0.56) or 90-day mortality (HR 1.13; 95% CI 0.87 to 1.47).
  • Serious adverse events (infection, arrhythmia, bleeding, seizures, metabolic/electrolyte disorders) were broadly similar between groups; no statistically significant harm signal was demonstrated for higher MAP targeting.

Internal Validity

  • Randomisation and allocation: Computer-generated assignment in permuted blocks (varying size), stratified by centre and initial rhythm; allocation concealment maintained via central randomisation and device-based masking.
  • Post-randomisation exclusions: 802 randomised; 789 included (12 withdrew consent for use of data; 1 randomised twice). Two patients were censored early (day 12 and day 13) due to incomplete follow-up data in non-residents.
  • Performance/detection bias: Blood-pressure intervention was double-blind using a calibrated module; this is unusually robust for haemodynamic target trials and limits co-intervention bias linked to knowledge of assignment.
  • Protocol adherence and separation: Early haemodynamic separation (2–48 hours) was substantial (mean MAP difference 10.5 mmHg) with higher vasoactive exposure; over the ICU stay, median MAP was 76 vs 72 mmHg, suggesting attenuation of separation beyond the early phase.
  • Baseline comparability: Groups were well balanced (e.g., age 63±13 vs 62±13 years; shockable rhythm 86% vs 84%; time to ROSC 21 (16–30) vs 21 (16–31) min; hypertension 38% vs 38%).
  • Outcome assessment: Primary outcome (death or CPC 3–4 at discharge) uses hard mortality plus a functional component; 3-month CPC/mRS were mainly via telephone interview, with some chart-based assessments when direct interview was not possible; MoCA and health-related quality of life had substantial missingness (COVID-era and other constraints).
  • Statistical rigour: Primary analysis used a stratified Cox model adjusted for factorial oxygenation assignment; intention-to-treat principle applied; interaction testing between BP and oxygenation targets prespecified and negative (P=0.89).

Conclusion on Internal Validity: Overall, internal validity appears moderate-to-strong given robust randomisation, unusually effective blinding for a haemodynamic target, and clear early separation, tempered by post-randomisation consent withdrawals, attenuation of separation over the ICU stay, and missingness in cognitive outcomes.

External Validity

  • Population representativeness: Participants were OHCA survivors with presumed cardiac cause and a high proportion of favourable arrest features (e.g., shockable rhythms, high bystander CPR rates), managed in specialised Danish cardiac arrest centres.
  • Important exclusions: Unwitnessed asystole, suspected non-cardiac aetiologies (e.g., intracranial bleed/intoxication), very early profound hypotension (MAP <50 mmHg), and delayed enrolment (>4 hours after ROSC) limit applicability to non-shockable, non-cardiac, or late-presenting cohorts.
  • Health-system context: High uptake of coronary angiography/PCI and protocolised post-arrest care may not translate to lower-resource systems; vasoactive practices (including dopamine use) may differ internationally.
  • Representativeness analysis: In a registry comparison from the trial documentation, trial participants differed from the broader OHCA-with-ROSC population at participating centres (e.g., shockable rhythm 85% vs 61%; ROSC at hospital arrival 77% vs 49%; 30-day mortality 30% vs 55%) 1

Conclusion on External Validity: Generalisability is moderate-to-limited: the findings best apply to comatose OHCA survivors of presumed cardiac cause treated in high-functioning cardiac arrest centres, and less directly to non-cardiac causes, non-shockable rhythms, in-hospital arrest, or resource-limited settings.

Strengths & Limitations

  • Strengths:
    • Methodological innovation: effective double blinding of a MAP-target intervention using a calibrated monitoring module.
    • Patient-centred primary endpoint incorporating mortality and severe functional outcome.
    • Early randomisation (median 146 minutes after ROSC) and protocolised post-arrest care pathways.
    • Factorial platform enabled assessment of interaction with oxygenation targets, reducing the risk that a bundled co-intervention explained findings.
  • Limitations:
    • Two-centre design with a selected, relatively favourable-prognosis OHCA population, limiting generalisability.
    • Achieved MAP separation was greatest early and smaller across the total ICU stay (median MAP 76 vs 72 mmHg), potentially limiting sensitivity to detect modest clinical effects.
    • Primary composite was largely driven by mortality; the disability component at discharge may be influenced by discharge practices even though assignment was masked.
    • Substantial missingness in cognitive and quality-of-life outcomes at 3 months, reducing interpretability for those endpoints.

Interpretation & Why It Matters

  • Clinical practice
    • Routine targeting of a higher MAP (around 77 mmHg) in unselected comatose OHCA survivors did not improve survival or functional outcome compared with a lower target (around 63 mmHg).
    • Given the increased vasopressor exposure required to achieve higher MAP, a default strategy of “avoid hypotension” (rather than “induce hypertension”) is supported for most patients.
  • Mechanistic inference
    • The absence of benefit despite early haemodynamic separation suggests that simply raising systemic MAP may not reliably translate into improved cerebral perfusion in post-arrest physiology (autoregulation impairment, microcirculatory dysfunction, and competing determinants of oxygen delivery).
  • Trial design implications
    • Future studies may need patient selection (e.g., chronic hypertension, autoregulation monitoring, shock phenotype) and/or more individualised perfusion targets rather than uniform MAP augmentation.

Controversies & Subsequent Evidence

  • The accompanying editorial emphasised that although blinding and early separation were strengths, higher MAP targeting required substantially more vasoactive support and did not translate into improved patient-centred outcomes 2
  • Prior randomised pilot work (e.g., COMACARE) was underpowered for clinical endpoints and often focused on biomarkers; BOX provided the first large-scale, blinded clinical-outcome test of higher vs lower MAP targets in this setting 3
  • An individual patient data meta-analysis pooling randomised trials comparing higher vs lower MAP targets after cardiac arrest reported no compelling evidence of benefit for higher targets on key clinical outcomes, aligning with BOX’s neutral results 4
  • Recent European expert consensus on intensive care after cardiac arrest highlights individualised haemodynamic management (avoid hypotension; consider patient phenotype and cerebral/organ perfusion monitoring where available) rather than uniform vasopressor-driven hypertension in all patients 5

Summary

  • In comatose survivors of presumed-cardiac OHCA, targeting MAP 77 mmHg did not reduce death or severe disability compared with targeting MAP 63 mmHg.
  • Early haemodynamic separation was achieved (mean MAP difference 10.5 mmHg from 2–48 h), but required greater vasopressor exposure.
  • Mortality at 90 days was similar between groups (31% vs 29%), and functional outcomes (CPC and mRS at 3 months) did not differ meaningfully.
  • Serious adverse events (infection, arrhythmia, bleeding, seizures, metabolic/electrolyte disorders) were broadly comparable between MAP targets.
  • BOX supports a default strategy of preventing hypotension rather than routinely pursuing higher-than-standard MAP targets in unselected post-arrest patients.

Further Reading

Other Trials

Systematic Review & Meta Analysis

Observational Studies

Guidelines

Notes

  • Where full bibliographic fields were not present in the available source material, “Not reported” is used rather than inferring volume/page details.
  • BOX should be interpreted alongside evolving evidence on personalised perfusion targets (e.g., autoregulation-guided MAP) and the broader post–cardiac arrest care bundle.

Overall Takeaway

BOX (blood-pressure arm) is a landmark because it tested a core, titratable haemodynamic decision—MAP targeting—using rigorous blinding and clinically meaningful endpoints in comatose OHCA survivors. Despite achieving early haemodynamic separation, higher MAP targeting did not improve survival or neurological outcomes, supporting a practice focus on avoiding hypotension and reserving aggressive MAP augmentation for selected phenotypes and future personalised trials.

Overall Summary

  • Higher MAP targeting (77 mmHg) increased vasoactive exposure but did not improve death or disability outcomes versus a lower target (63 mmHg).
  • Blinded haemodynamic target delivery is feasible and materially strengthens causal inference in post-arrest bundle trials.
  • Subsequent syntheses and expert consensus align with BOX: prevent hypotension; individualise when physiology suggests benefit.

Bibliography