Predictive utility of weekly post-traumatic amnesia assessments after brain injury: A multicentre analysis

Allen W. Brown, James F. Malec, Jay Mandrekar, Nancy N. Diehl, Sureyya S. Dikmen, Mark Sherer, Tessa Hart, Thomas A. Novack

Research output: Contribution to journalArticlepeer-review

21 Scopus citations


Primary objective: To consider the duration of post-traumatic amnesia (PTA) as a single predictor variable to determine whether categories of PTA duration exist that predict 1-year outcomes after traumatic brain injury (TBI). Research design: Using the TBI Model System database (n 5250), the duration of PTA that predicts 1-year outcomes was calculated. Logistic regression was used to determine whether a single predictive threshold value existed. Classification and regression tree analysis then determined whether multiple threshold values existed. The area under the ROC curve and percentage correct classification were also calculated to discriminate classification accuracy and choose the best predictive thresholds. Results: A single threshold value of PTA days was identified for all variables. At 1 year, PTA that extended into week 7 after injury predicted total FIM™ and FIM™ sub-scales. One-year employment and global outcomes were predicted by PTA extending into week 4 after injury. Independent living was predicted by PTA extending into the 8th week after injury. Conclusion: After moderate-to-severe TBI, determining the presence of PTA at weekly intervals can efficiently determine injury severity while providing meaningful precision when used in research and clinical outcome prediction.

Original languageEnglish (US)
Pages (from-to)472-478
Number of pages7
JournalBrain Injury
Issue number3
StatePublished - 2010


  • Measurement
  • Outcome
  • Rehabilitation

ASJC Scopus subject areas

  • Neuroscience (miscellaneous)
  • Developmental and Educational Psychology
  • Clinical Neurology


Dive into the research topics of 'Predictive utility of weekly post-traumatic amnesia assessments after brain injury: A multicentre analysis'. Together they form a unique fingerprint.

Cite this