Identifying Patterns of Medical Intervention in Acute Respiratory Failure: A Retrospective Observational Study.

Area of Science:

• Critical care medicine
• Health services research
• Data science in healthcare

Background:

• Understanding patient trajectories in intensive care units (ICUs) is crucial for managing expectations and informing clinical decisions.
• Characterizing medical interventions over time can reveal patterns related to patient outcomes.

Purpose of the Study:

• To identify distinct and clinically relevant pathways of medical intervention among adult ICU patients diagnosed with acute respiratory failure.
• To explore the relationship between these intervention pathways and patient outcomes.

Main Methods:

• A retrospective observational study utilizing administrative claims data from 2012 to 2014.
• Temporal k-means clustering was applied to patterns of cumulative medical intervention delivery at specific hospital days (0, 5, 10, 20, and up to discharge).
• The study included 12,175 admissions from four states, divided into training and validation sets.

Main Results:

• Three distinct intervention pathways were identified, accounting for 93.5% of admissions: 'cardiac' (45.9%), 'general' (36.7%), and 'prolonged' (17.4%).
• The 'prolonged' pathway was associated with significantly longer hospital stays (median 13 days) and higher in-hospital mortality (24.6%) compared to 'cardiac' and 'general' pathways.
• These findings were validated in a separate dataset.

Conclusions:

• The majority of ICU admissions for acute respiratory failure can be categorized into one of three clinically meaningful intervention pathways.
• These pathways are demonstrably linked to hospitalization outcomes, including length of stay and mortality.
• This classification provides a framework for predicting patient outcomes, improving patient-family communication, and optimizing critical care resource allocation.