The Neurosurgery Signal — Signal, Not Noise

Rapid review note

Journal Club is a rapid, AI-assisted appraisal layer. It highlights study design, effect estimates, and practice relevance, but it is still a briefing, not a replacement for the paper.

For education only. Not medical advice.

Paper snapshot

Rapid study overview

Open paper

DOI

10.1001/jamaneurol.2020.3745

PMID

N/A

PICO

Population

Patients with space-occupying hemispheric infarction from 7 randomized trials in 6 countries

Intervention

Surgical decompression (hemicraniectomy with duraplasty)

Comparator

Medical treatment alone

Outcomes

Favorable functional outcome, death, and mRS shift at 6 months and 1 year

Design

Type

Systematic review with individual patient data meta-analysis of randomized trials

Randomized

Multicenter

Yes

Blinded

mixed across included trials

Follow-up

Primary endpoint at 1 year after stroke

Primary endpoint

Favorable outcome defined as mRS <=3 at 1 year

Secondary endpoints

Death at 6 months and 1 year
Reasonable outcome mRS <=4
Excellent outcome mRS <=2
Ordinal mRS shift

Practice impact

What this means

This individual patient meta-analysis of randomized trials supports decompressive surgery for malignant hemispheric infarction, with large mortality benefit and better chances of mRS <=3 at 1 year. Apply caution for late presenters and elderly subgroup extrapolation.

Bottom line

Evidence supports decompressive surgery to reduce mortality and improve odds of favorable function in malignant hemispheric infarction.

Strength of evidence

high

Recommendation

change practice

Why it matters

Large effect size for mortality reduction across pooled randomized data.
Functional outcome odds also improved.
Subgroup effects were generally consistent in prespecified analyses.

What would change my mind

New randomized data showing no net functional benefit in modern ICU practice.
Robust late-window randomized evidence contradicting early-treatment effects.
Patient-centered quality-of-life data showing unacceptable tradeoffs in key subgroups.

Critical appraisal

How strong is the paper?

Methods critique

Risk of bias

Most included trials were low to moderate risk, but one had high risk and trial-level methods varied.

Confounding

Randomization in source trials limits confounding, but pooled analysis still depends on trial-level protocol differences.

Missing data

Some source trial data were incomplete or unavailable, and one unpublished trial contributed limited data.

Multiplicity

Multiple secondary and subgroup analyses increase false-positive risk, though prespecified plans were used.

Notes

Few patients were randomized after 48 hours, limiting late-window inference.
Elderly subgroup estimates varied across studies.

Stats check

NNT

NNT about 5 for one additional mRS <=3 outcome at 1 year from RD 21%

Effect sizes

Adjusted OR for death: 0.16 (95% CI 0.10-0.24)
Adjusted OR for favorable mRS <=3 at 1 year: 2.95 (95% CI 1.55-5.60)
No heterogeneity signal across prespecified subgroups

Absolute effects

Risk difference for mRS <=3 at 1 year: 21% (95% CI 9-33)
Direction of effect favored surgery for mortality and functional outcomes at 6 and 12 months

Concerns

Absolute event rates differ by trial and era.
Limited late-treatment sample weakens conclusions beyond 48 hours.

External validity

Who it applies to

Adults with malignant hemispheric infarction considered for decompressive surgery in monitored stroke centers.

Who it does not

Patients presenting late beyond evidence-supported windows or with major contraindications to surgery.

Generalizability notes

Multinational randomized datasets improve transportability.
Older-age conclusions are less certain because subgroup counts were small.

Evidence trace

Source trace and metadata

Citations (5)

claim_id

methods_critique.risk_of_bias

locator

p. 1 Abstract Results

claim_id

methods_critique.missing_data

locator

p. 3 Results

claim_id

stats_check.effect_death

locator

p. 1 Abstract Results

claim_id

stats_check.primary_rd

locator

Table 3

claim_id

practice_impact.bottom_line

locator

p. 1 Conclusions and Relevance

Metadata

Generated at

2026-03-04T03:09:55Z

Version