Skip to main content

Foundational Trials

STICH

Image: Shutterstock / ChaNaWiT

Publication

  • Title: Early surgery versus initial conservative treatment in patients with spontaneous supratentorial intracerebral haematomas in the International Surgical Trial in Intracerebral Haemorrhage (STICH): a randomised trial
  • Acronym: STICH
  • Year: 2005
  • Journal published in: The Lancet
  • Citation: Mendelow AD, Gregson BA, Fernandes HM, Murray GD, Teasdale GM, Hope DT, Karimi A, Shaw MDM, Barer DH, for the STICH investigators. Early surgery versus initial conservative treatment in patients with spontaneous supratentorial intracerebral haematomas in the International Surgical Trial in Intracerebral Haemorrhage (STICH): a randomised trial. Lancet. 2005;365(9457):387-97.

Context & Rationale

  • Background
    • Spontaneous intracerebral haemorrhage (ICH) carries high early mortality and substantial long-term disability, with limited disease-modifying therapies.
    • Open haematoma evacuation was widely practised but clinically contentious, with prior randomised evidence small, heterogeneous, and inconclusive.
    • Theoretical benefits of surgery included reduction in mass effect/raised intracranial pressure and potential mitigation of secondary injury; theoretical harms included surgical trauma and perioperative rebleeding.
    • Practice variation and uncertainty about which anatomical/clinical subgroups might benefit were major drivers for a definitive pragmatic trial.
  • Research Question/Hypothesis
    • In patients with CT-confirmed spontaneous supratentorial ICH for whom a neurosurgeon was uncertain about best management, does a policy of early surgery (haematoma evacuation within 24 hours of randomisation) improve functional outcome compared with initial conservative treatment?
    • Hypothesis: early evacuation would reduce death and disability at 6 months compared with initial conservative management.
  • Why This Matters
    • ICH frequently presents to emergency/critical care pathways, and early neurosurgical triage decisions have major resource and outcome implications.
    • A neutral or negative result would challenge routine early craniotomy policies; a positive result would justify system-wide pathways for rapid transfer and operative capability.
    • Clarifying subgroup effects could inform precision selection (e.g., superficial lobar haemorrhage versus deep basal ganglia haemorrhage) and guide subsequent trial design.

Design & Methods

  • Research Question: Among patients with spontaneous supratentorial ICH where the treating neurosurgeon is uncertain about management, does a policy of early surgical haematoma evacuation (within 24 hours of randomisation) improve 6-month functional outcome compared with initial conservative treatment?
  • Study Type: International, multicentre, randomised, pragmatic, parallel-group trial conducted in neurosurgical units (83 centres in 27 countries); open-label treatment with structured 6-month outcome collection.
  • Population:
    • Setting: neurosurgical units with CT-confirmed diagnosis.
    • Inclusion: spontaneous supratentorial intracerebral haematoma confirmed by CT; randomisation within 72 hours of ictus; “clinical uncertainty principle” (surgeon uncertain whether early surgery or conservative care is preferable).
    • Operational recommendations: minimum haematoma diameter 2 cm; Glasgow Coma Score (GCS) ≥5 (recommendations; eligibility governed by uncertainty).
    • Exclusion: probable aneurysm/arteriovenous malformation; haemorrhage secondary to tumour or trauma; cerebellar haemorrhage or extension into the brainstem; severe pre-existing physical/mental disability or severe comorbidity expected to interfere with assessment; inability to undertake surgery within 24 hours of randomisation.
  • Intervention:
    • Early surgery policy: evacuation of the intracerebral haematoma within 24 hours of randomisation, plus best medical management.
    • Operative technique: at surgeon discretion (craniotomy predominant; burr hole, endoscopy, stereotaxy, and other methods permitted).
    • Perioperative/ICU co-interventions: not standardised; delivered according to local practice (Not reported in detail).
  • Comparison:
    • Initial conservative policy: best medical management without immediate surgery.
    • Rescue surgery: permitted if subsequently judged necessary (e.g., neurological deterioration or other clinical indications).
    • Co-interventions: not standardised; local practice (Not reported in detail).
  • Blinding: Treatment allocation was not blinded (surgery versus no immediate surgery); 6-month outcomes were collected using a postal questionnaire approach (reducing but not eliminating detection bias given patient awareness of treatment).
  • Statistics: Sample size assumed 40% favourable outcome with initial conservative treatment; a 10% absolute improvement was targeted; with alpha 0.05 and 80% power, 800 patients were required; inflated by 25% to 1000 to allow for protocol violations; primary analysis was intention-to-treat (favourable vs unfavourable outcome at 6 months using a prognosis-based Glasgow Outcome Scale definition).
  • Follow-Up Period: 6 months for primary functional outcome and mortality assessment.

Key Results

This trial was not stopped early. Recruitment and follow-up proceeded to the prespecified target sample size.

Outcome Early surgery policy Initial conservative policy Effect p value / 95% CI Notes
Primary: favourable outcome at 6 months (prognosis-based Glasgow Outcome Scale) 122/468 (26%) 118/496 (24%) OR 0.89 95% CI 0.66 to 1.19; P=0.414 Favourable outcome defined by a “sliding dichotomy” using a prognostic score; absolute benefit 2.3% (−3.2 to 7.7).
Mortality at 6 months 170/468 (36%) 184/496 (37%) OR 0.95 95% CI 0.73 to 1.23; P=0.707 Absolute benefit 0.7% (−5.0 to 6.4).
Secondary: favourable modified Rankin Scale at 6 months 152/468 (33%) 137/496 (28%) OR 0.82 95% CI 0.62 to 1.09; P=0.116 Absolute benefit 4.7% (−1.1 to 10.5).
Secondary: favourable Barthel Index at 6 months 124/468 (27%) 110/496 (23%) OR 0.84 95% CI 0.63 to 1.12; P=0.144 Absolute benefit 4.1% (−1.5 to 9.7).
Prespecified subgroup: unfavourable outcome (prognosis-based) when haematoma ≤1 cm from cortical surface 158/196 (81%) 208/265 (79%) RR 0.83 95% CI 0.73 to 0.96; P=0.02 Direction favoured early surgery in superficial haemorrhage; absolute benefit 8% (0 to 15).
Prespecified subgroup: unfavourable outcome (prognosis-based) when haematoma >1 cm from cortical surface 247/300 (82%) 238/323 (74%) RR 1.04 95% CI 0.92 to 1.18; P=0.59 No evidence of benefit for deeper haemorrhage; absolute benefit −2% (−10 to 6).
Prespecified subgroup: unfavourable outcome (prognosis-based) when comatose at presentation (GCS 3–8) 340/450 (76%) 322/434 (74%) RR 1.08 95% CI 0.99 to 1.18; P=0.08 Trend towards worse outcome with early surgery in coma; absolute benefit −6% (−14 to 2).
  • Across the whole randomised population, early surgery did not improve 6-month functional outcome or survival compared with an initial conservative policy.
  • Treatment separation was substantial but imperfect: 94% of those allocated early surgery underwent evacuation, while 26% of those allocated conservative care later underwent surgery after deterioration (biasing towards a null effect).
  • A prespecified anatomical subgroup (haematoma ≤1 cm from the cortical surface) showed a signal consistent with benefit of early surgery, whereas deeper haemorrhages did not.

Internal Validity

  • Randomisation and Allocation: Central 24-hour telephone randomisation with minimisation; allocation concealment at enrolment was strong.
  • Drop out / exclusions: At 6 months, 51 patients were lost to follow-up and 17 were alive but their status was unknown; the primary outcome analysis included 468/503 (93%) in the early-surgery group and 496/530 (94%) in the conservative group (missing data ~7% overall).
  • Performance/Detection Bias: Open-label treatment created risk of performance bias; 6-month outcomes were collected using postal questionnaires, reducing clinician-assessor influence but not removing patient/carer expectation effects.
  • Protocol Adherence: Early-surgery delivery was high (465/503 [94%] underwent surgery; 28/465 [6%] had surgery >24 hours after randomisation; 25/503 [~5%] did not undergo surgery).
  • Baseline Characteristics: Groups were broadly comparable (median age 61 vs 62 years; median haematoma volume 30 mL vs 30 mL; median time ictus-to-randomisation 22 h vs 21 h; GCS 9–12: 45% vs 44%).
  • Heterogeneity: Pragmatic multinational design enhanced representativeness but increased between-centre variability; surgical technique was not standardised (craniotomy 75% vs 85% among operated patients; endoscopy 7% vs 5%; stereotaxy 7% vs 2%).
  • Timing: The intervention was early relative to randomisation (median 5 h from randomisation to surgery [IQR 2–12] versus 31 h [11–82] in those later operated after initial conservative care), but not “ultra-early” from ictus (median 30 h [16–49] versus 60 h [27–99]).
  • Dose: Extent/completeness of haematoma evacuation and residual clot volume targets were not standardised (Not reported), limiting inference about a “biological dose–response”.
  • Separation of the Variable of Interest: Surgery rate 94% (early policy) versus 26% (initial conservative policy); surgery within 12 hours of randomisation 339/465 (73%) versus 35/140 (25%); median randomisation-to-surgery 5 h versus 31 h.
  • Crossover: 140/530 (26%) allocated to conservative care underwent surgery later, usually after neurological deterioration; the operated conservative subgroup was substantially worse at surgery (GCS 3–8: 73% at surgery versus 24% in early-surgery operations).
  • Outcome Assessment: Primary endpoint used a prognosis-based dichotomy (sliding dichotomy) applied to the extended Glasgow Outcome Scale; secondary outcomes used mRS and Barthel Index.
  • Statistical Rigor: Prespecified power target was achieved (1033 randomised); primary analysis was intention-to-treat with prespecified subgroup interaction testing.

Conclusion on Internal Validity: Overall internal validity is moderate: randomisation and prespecified analyses were robust, but open-label delivery, treatment crossover (26% in the conservative group receiving surgery), and heterogeneity in surgical technique/timing plausibly diluted any true subgroup-specific effect.

External Validity

  • Population Representativeness: Broad international recruitment (83 centres, 27 countries) supports generalisability to neurosurgical-unit patients with supratentorial ICH where genuine equipoise exists.
  • Key exclusions limiting generalisability: Secondary haemorrhage (aneurysm/AVM/tumour/trauma), cerebellar haemorrhage, brainstem extension, severe pre-morbid disability, and patients in whom early surgery could not be delivered within 24 hours.
  • Clinical “equipoise” enrolment: The “clinical uncertainty principle” enriches for borderline cases and reduces applicability to patients with clear surgical indications (or clear futility), which are common in real-world ICU triage.
  • Applicability across systems: Findings apply best where rapid transfer to neurosurgery is feasible; health systems with delayed access may have poorer alignment with the trial’s timing.
  • Technique evolution: Predominantly open craniotomy-era evidence may not translate directly to modern minimally invasive or image-guided clot-reduction strategies.

Conclusion on External Validity: External validity is moderate: highly relevant to selected supratentorial ICH patients managed in neurosurgical units with equipoise, but less applicable to patients with definitive surgical indications/contraindications, delayed presentation/transfer, or modern minimally invasive pathways.

Strengths & Limitations

  • Strengths:
    • Largest randomised evaluation at the time of early surgery versus initial conservative care for supratentorial ICH.
    • International, multicentre pragmatic design capturing real-world neurosurgical practice variability.
    • Central randomisation with minimisation and intention-to-treat analysis.
    • Use of a prognosis-based (“sliding dichotomy”) primary endpoint intended to improve sensitivity across severity strata.
  • Limitations:
    • Open-label intervention with non-standardised co-interventions and surgical technique.
    • Substantial crossover (26% of conservative group underwent delayed surgery), likely biasing effects towards the null.
    • Median ictus-to-surgery timing (~30 h) may have been too late for some mechanistic hypotheses of early tissue salvage.
    • Primary endpoint complexity (prognosis-based dichotomy) reduces bedside interpretability and invites sensitivity to modelling choices.
    • Exclusion of cerebellar haemorrhage and secondary causes limits direct applicability to several high-stakes surgical scenarios.

Interpretation & Why It Matters

  • Routine early craniotomy
    STICH does not support a blanket policy of early open evacuation for all spontaneous supratentorial ICH patients who reach neurosurgical services (primary favourable outcome 26% vs 24%; OR 0.89; 95% CI 0.66 to 1.19; P=0.414).
  • Patient selection is pivotal
    The prespecified anatomical signal in superficial haemorrhage (≤1 cm from cortex: RR 0.83 for unfavourable outcome; 95% CI 0.73 to 0.96; P=0.02) argues for targeted selection rather than universal early surgery.
  • ICU pathway implications
    For critical care clinicians, the trial reframes early neurosurgical referral as a selection-and-timing problem (who, how quickly, and by what technique), rather than default transfer-to-craniotomy for all supratentorial ICH.

Controversies & Subsequent Evidence

  • Interpretation of a “neutral” overall result: The accompanying editorial emphasised that STICH addressed a long-standing controversy, highlighted the superficial-haematoma signal, and argued that deep haemorrhage might require different (less invasive) surgical strategies rather than open craniotomy.1
  • Prognosis-based (“sliding dichotomy”) endpoint: The primary endpoint’s dependence on a prognostic model was methodologically purposeful but less intuitive for clinicians; the approach and its implications were described by the investigators in a dedicated methods paper.2
  • Equipoise-driven enrolment and selection effects: The “clinical uncertainty principle” improves ethical conduct but limits applicability to cases with clear surgical indications; an investigator report documented major international variation in surgical practice surrounding STICH, reinforcing both heterogeneity and selection concerns.3
  • Follow-up trial refining the target population: STICH II focused on spontaneous superficial lobar ICH (a subgroup suggested by STICH) and again found no statistically significant overall benefit of early surgery, although the trial was designed to reduce dilution from deep haemorrhage.4
  • Guideline incorporation: Contemporary stroke guidelines incorporate STICH/STICH II by discouraging routine supratentorial evacuation as a default strategy while supporting surgery in selected scenarios (e.g., life-threatening mass effect or neurological deterioration) and strongly endorsing urgent surgery for cerebellar haemorrhage (a population excluded from STICH).5

Summary

  • STICH randomised 1033 patients with spontaneous supratentorial ICH (within 72 h) where neurosurgeons were uncertain about best management, to early surgery within 24 h versus initial conservative care.
  • There was no significant difference in the primary 6-month prognosis-based functional outcome (26% vs 24%; OR 0.89; 95% CI 0.66 to 1.19; P=0.414) or 6-month mortality (36% vs 37%; OR 0.95; 95% CI 0.73 to 1.23; P=0.707).
  • Treatment crossover was substantial: 26% of the conservative group underwent delayed surgery, and those delayed patients were markedly more neurologically compromised at the time of surgery.
  • A prespecified subgroup suggested possible benefit in superficial haemorrhage (≤1 cm from cortex: RR 0.83 for unfavourable outcome; 95% CI 0.73 to 0.96; P=0.02), whereas deeper haemorrhages showed no benefit.
  • The trial established that routine early open evacuation is not universally beneficial and redirected the field towards patient selection, timing, and technique (including minimally invasive approaches).

Further Reading

Other Trials

Systematic Review & Meta Analysis

Observational Studies

Guidelines

Notes

  • Where a DOI/journal landing page could not be verified from accessible metadata for older studies, a PubMed search link is provided to the published literature record(s).

Overall Takeaway

STICH was a landmark pragmatic international trial demonstrating that a routine policy of early open haematoma evacuation for spontaneous supratentorial ICH does not improve overall 6-month functional outcome or survival compared with initial conservative care. Its enduring contribution is to shift neurosurgical and critical care practice away from “one-size-fits-all” craniotomy and towards evidence-based selection (particularly superficial lobar haemorrhage), timing, and technique refinement.

Overall Summary

  • Early open evacuation did not improve overall 6-month outcome or mortality versus initial conservative care in equipoise-selected supratentorial ICH.
  • High crossover (26% delayed surgery in the conservative group) and heterogeneous techniques likely diluted any subgroup-specific benefit.
  • A prespecified signal in superficial haemorrhage catalysed later targeted trials and informs contemporary guideline nuance.

Bibliography