Decompressive craniectomy for stroke: decision tree helps predict tracheostomy need

Access to the full content of this site is available only to registered healthcare professionals. Register to read more


  • A new decision tree using preoperative clinical data had good performance for predicting tracheostomy need in patients undergoing decompressive craniectomy for stroke.

Why this matters

  • Tracheostomy placement may shorten duration of mechanical ventilation in patients undergoing decompressive craniectomy for stroke.
  • Optimal timing of the procedure is unclear.

Key results

  • Overall, 28.5% of the patients needed tracheostomy.
  • A decision tree using Glasgow Coma Scale, Sequential Organ Failure Assessment score, and presence of hydrocephalus (defined as any neurologic deterioration attributable to elevated intracranial pressure) had 63% sensitivity and 84% specificity for predicting tracheostomy need.
  • Compared with patients undergoing tracheostomy later (>10 days after stroke), those undergoing tracheostomy early (≤10 days) had fewer ventilator days (7.3 vs 15.2 days; P<.001) and shorter hospital length of stay (28.5 vs 44.4 days; P=.014), but rates of ventilator-associated pneumonia and mortality were similar.

Study design

  • A retrospective analysis of prospectively collected registry data for 168 adult patients who underwent decompressive craniectomy for ischemic stroke or intracerebral hemorrhage.
  • Main outcome was predicted need for tracheostomy using a propensity-weighted decision tree analysis built with factors present before surgical decompression.
  • Funding: Author-provided funding.


  • The study was fairly small and retrospective.
  • Findings may have been affected by residual confounding.