Cardiovascular

Trajectory-Based Modeling Improves Long-Term Outcome Prediction Beyond 90-Day Assessment in ICU-Treated Intracerebral Hemorrhage.

TL;DR

Trajectory-based classification of functional recovery in ICU-treated intracerebral hemorrhage patients provides superior predictive performance for 12-month quality of life compared with conventional 90-day mRS assessment (AUC 0.81 vs 0.72; p = 0.003), supporting a shift toward dynamic outcome modeling for long-term prognostication in ICH.

Key Findings

Three distinct functional recovery trajectories were identified among ICU-treated ICH survivors using latent class mixed modeling.

  • The study enrolled 196 patients with a mean age of 61.7 ± 12.4 years, of whom 60.2% were male.
  • Patients were admitted to a tertiary ICU between January 2024 and January 2025 and followed for 12 months post-discharge.
  • The three trajectories were classified as good, moderate, and poor recovery.
  • Functional status was assessed using the modified Rankin Scale (mRS) at hospital discharge, 90 days, 6 months, and 12 months.

Twelve-month health-related quality of life demonstrated a graded decline across the three trajectory groups.

  • QoL at 12 months was measured using the EuroQol 5-Dimension 5-Level questionnaire (EQ-5D-5L).
  • The graded decline in QoL followed the order of good, moderate, and poor recovery trajectory groups.
  • This association supported the clinical validity of the trajectory classification system.

Older age, lower admission Glasgow Coma Scale scores, and larger hematoma volume were independently associated with both moderate and poor recovery trajectories.

  • These associations were identified via multinomial logistic regression evaluating early clinical characteristics and trajectory membership.
  • These three factors were common predictors shared across both the moderate and poor recovery trajectory groups.
  • The analysis evaluated multiple early clinical characteristics as candidate predictors of trajectory membership.

Intraventricular hemorrhage, mechanical ventilation, and prolonged sedation (>48 hours) were specifically associated with the poor recovery trajectory.

  • These three factors were distinct from the predictors shared with the moderate recovery trajectory.
  • Prolonged sedation was defined as sedation duration exceeding 48 hours.
  • These ICU-specific variables were identified through multinomial logistic regression as trajectory membership predictors.

Trajectory-based classification demonstrated superior predictive performance for 12-month quality of life compared with 90-day mRS.

  • The area under the curve (AUC) for trajectory-based classification was 0.81 versus 0.72 for 90-day mRS (p = 0.003).
  • Trajectory-based classification improved risk reclassification with a net reclassification improvement (NRI) of 0.27.
  • The integrated discrimination improvement (IDI) was 0.09, indicating significant improvement in risk discrimination.
  • Receiver operating characteristic curve analysis was used to compare the predictive performance of the two approaches.

What This Means

This research suggests that tracking how brain bleed (intracerebral hemorrhage) patients recover over time — rather than taking a single snapshot at 90 days — gives doctors a much better picture of how those patients will fare a year later. The study followed 196 patients who were treated in an intensive care unit after suffering a spontaneous brain bleed, assessing their physical functioning at hospital discharge, 3 months, 6 months, and 12 months. Using a statistical technique called latent class mixed modeling, researchers found that patients naturally grouped into three distinct recovery patterns: good, moderate, and poor recovery, each linked to meaningfully different quality of life one year later. The study also identified which early characteristics predicted which recovery path a patient would follow. Factors like older age, lower consciousness levels at admission, and larger bleeds predicted both moderate and poor outcomes. However, complications such as bleeding into the brain's ventricles, needing a breathing machine, and being kept sedated for more than 48 hours were specifically tied to the worst recovery group. Importantly, the trajectory-based approach predicted 12-month quality of life significantly better than the standard single-point 90-day assessment, with the trajectory model achieving an accuracy score (AUC) of 0.81 compared to 0.72 for the conventional approach. This research suggests that following patients' recovery as a pattern over time — rather than relying on a single assessment — could help clinicians and families better understand long-term outcomes after serious brain bleeds. The findings support moving toward more dynamic, time-based methods of predicting recovery for ICU patients with intracerebral hemorrhage, which could ultimately inform rehabilitation planning and resource allocation.

Have a question about this study?

Citation

Su Q, Zhang X, Wang B, Zhu L, Li Y. (2026). Trajectory-Based Modeling Improves Long-Term Outcome Prediction Beyond 90-Day Assessment in ICU-Treated Intracerebral Hemorrhage.. Revista de neurologia. https://doi.org/10.31083/RN52441