Generation and Validation of a Clinical Prediction Rule for Sedative-Associated Delirium During Acute Respiratory Failure - PROJECT SUMMARY/ABSTRACT
More than half of the >10 million patients who are mechanically ventilated for acute respiratory failure
throughout the world each year experience delirium, a brain dysfunction syndrome characterized by acute
disturbances in attention, awareness, and cognition. Delirium is a risk factor for numerous adverse outcomes,
including delayed liberation from mechanical ventilation, increased health care costs, increased mortality,
and—for those who survive—long-term disability and persistent cognitive impairment. Up to one third of
survivors of acute respiratory failure are affected by long-term cognitive impairment, the severity of which is
similar to that caused by traumatic brain injury or Alzheimer’s Disease. The pathophysiology of delirium during
acute respiratory failure remains poorly understood, and effective therapies have yet to be identified. Thus,
avoidance of risk factors is key to improving outcomes. One of the common risk factors for delirium during
acute respiratory failure is exposure to sedating medications, which are often necessary when managing
mechanically ventilated patients. When choosing whether and how to sedate a patient with acute respiratory
failure, a clinician would ideally use information about risk for sedative-associated delirium, the duration of
which predicts long-term cognitive impairment. The identification and quantification of risk for sedative-
associated delirium during acute respiratory failure would allow clinicians to personalize decision-making
regarding sedation in the intensive care unit (ICU), making efforts to avoid sedation (or avoid specific
sedatives) when managing those patients at high risk for sedative-associated delirium. Such an approach to
personalized sedation would lead to less delirium and to improved patient-centered outcomes. Thus, we plan
in this NRSA project to test the central hypothesis that risk for sedative-associated delirium during acute
respiratory failure can be quantified in a clinical prediction rule based on patient characteristics and treatment
factors. To test this hypothesis, we will accomplish two specific aims. First, we use existing data collected
during the multicenter, prospective BRAIN-ICU cohort study to derive a clinical prediction rule that uses readily
available patient characteristics and treatment factors to predict sedative-associated delirium among adults
with acute respiratory failure (Aim 1). Next, we will validate the sedative-associated delirium clinical prediction
rule both internally to correct estimates for optimism (Aim 2A) and externally, assessing calibration and
discrimination in two ongoing, NHLBI-funded prospective cohort studies (Aim 2B). During this project, the
applicant’s mentors and training program will provide him with expert training in clinical research methodology
and cognitive assessments and outcomes, and he will cultivate the skills necessary for a career as an
independent clinical researcher in the field of critical care.
