← Back

ECASS I Post Hoc

Dichotomized Efficacy End Points and Global End-Point Analysis Applied to the ECASS Intention-to-Treat Data Set: Post Hoc Analysis of ECASS I

Share Share Copied! Compare

Year of Publication: 1998

Authors: Werner Hacke, Erich Bluhmki, Thorsten Steiner, ..., Dieter Meier

Journal: Stroke

Citation: Stroke. 1998;29:2073–2075

Link: https://www.ahajournals.org/doi/full/10.....STR.29.10.2073

PDF: https://www.ahajournals.org/doi/reader/1....STR.29.10.2073

Clinical Question

Would applying the global end-point analysis from the NINDS trial reveal a significant benefit for rtPA in the original ECASS I trial?

Bottom Line

Post hoc reanalysis of ECASS I ITT data (615 patients) using NINDS global endpoint methodology converted the 'negative' trial to 'positive': global OR 1.5 (95% CI 1.1-2.0; P=0.008). Individual outcomes: mRS 0-1 36% vs 28% (OR 1.4; P=0.044); NIHSS 0-1 36% vs 22% (OR 1.9; P=0.001); Barthel ≥95 44% vs 38% (OR 1.3; P=0.102, NS). Suggests benefit of IV rtPA (1.1 mg/kg) extends to 6 hours, but is a post hoc non-prespecified analysis.

        Major Points
        ECASS I ITT data reanalyzed with NINDS global endpoint method: global OR 1.5 (95% CI 1.1-2.0; P=0.008).
NIHSS 0-1: 36% vs 22% (OR 1.9; P=0.001) — largest individual effect.
mRS 0-1: 36% vs 28% (OR 1.4; P=0.044).
Barthel ≥95: 44% vs 38% (OR 1.3; P=0.102, NS).
Effect sizes smaller than NINDS for mRS (+8% vs +13%) and BI (+6% vs +12%), consistent with diminishing benefit at longer time-to-treatment.
rtPA dose was 1.1 mg/kg (higher than NINDS 0.9 mg/kg) within a 6-hour window.
Demonstrates critical importance of pre-specifying endpoints — same data yields opposite conclusions depending on analysis method.
Post hoc, non-prespecified analysis — cannot override the original negative ECASS I result.
Provided methodological justification for ECASS II and ECASS III design.
615 patients, 6-hour window. Funded by Boehringer Ingelheim.

      

Design

Study Type: Post hoc reanalysis of a randomized controlled trial

Randomization: 1

Blinding: Not specified for reanalysis

Enrollment Period: Original ECASS: 1992–1994

Follow-up Duration: 90 days

Countries: Europe

Sample Size: 615

Analysis: Global end-point analysis using generalized estimating equations; individual dichotomized outcomes assessed by Fisher's exact test

Inclusion Criteria

Acute ischemic hemispheric stroke
Treatment within 6 hours of onset
Age 18–80
CT scan excluding hemorrhage and major early infarct signs

Exclusion Criteria

Evidence of hemorrhage on baseline CT
Major CT protocol violations (e.g., extensive infarction)
Coma or very severe stroke (SSS <10)

Baseline Characteristics

Characteristic	Comorbidities	Qualifying Event

Arms

Field	rtPA	Control
Intervention	Recombinant tissue plasminogen activator (1.1 mg/kg) within 6 hours	Placebo infusion under identical conditions
Duration	Single IV infusion	Single IV infusion

Outcomes

Outcome	Type	Control	Intervention	HR / OR / RR	P-value
Global outcome based on 3 dichotomized endpoints: mRS 0–1, BI 95–100, NIHSS 0–1	Primary	28% (mRS), 38% (BI), 22% (NIHSS)	36% (mRS), 44% (BI), 36% (NIHSS)	8.00%	0.008
mRS 0–1 at 90 days \| 95% CI: 1.0–2.0; NNT: 13	Secondary	28%	36%		0.044
BI 95–100 at 90 days \| 95% CI: 0.9–1.8	Secondary	38%	44%		0.102
NIHSS 0–1 at 90 days \| 95% CI: 1.4–2.8; NNT: 8	Secondary	22%	36%		0.001
Mortality at 90 days	Adverse	16%	Higher (exact % not specified)
Protocol Violations (CT)	Adverse		Major CT violations affected trial results; excluded in per-protocol target analysis

Subgroup Analysis

Target population (per protocol) showed significant mRS difference (P=0.035); however, analysis in full ITT sample only reached significance via global test [oai_citation:1‡ECASS I Post hoc.pdf](file-service://file-7WvDM13r3CmnMS8rMjg6sb)

Criticisms

Post hoc analysis not predefined in original trial design
Global end-point strategy may inflate positive findings via redundant measures
BI did not reach statistical significance individually (P=0.102)
Increased mortality in rtPA group vs. placebo (likely due to CT protocol violations)
No formal adjustment for multiplicity

Funding

Boehringer Ingelheim (sponsor of ECASS I); authors include sponsor-employed statisticians

Based on: ECASS I Post Hoc (Stroke, 1998)