Acute lung injury (ALI) affects approximately 190,000 patients a year and is associated with more deaths annually in the United States than either breast cancer or HIV/AIDS. Clinical research in ALI is challenging because clinically important events recorded through time are complex and difficult to analyze. Physicians make decisions about the care of critically ill patients utilizing clinical information gathered through time; however, evaluations of treatments in critical care studies commonly use outcomes evaluated at 28 days such as hospital mortality and ventilator-free days. These outcomes are static in that they provide no detail concerning the effects of an intervention through time. When the effects through time are ignored, important information on valuable morbidity outcomes is lost. A shorter time on the ventilator or in the intensive care unit is of significance to patients because they may decrease the chance of adverse events or the need for more intensive care procedures. Our first objective is to fully characterize the effects of interventions on important morbidity outcomes through time. To achieve this objective, we will use a novel statistical approach to model two clinically-meaningful outcomes jointly through time. Patient-related factors can also be time-dependent and their sequence and duration can affect the occurrence of clinical outcomes. Our second objective is to determine the association between time-varying factors and the evolution of timed outcomes. There is potentially great value in evaluating longitudinally collected information, such as repeated measurements of plateau pressure or the temporal sequence of organ failures; however, longitudinal (time-varying) data have been largely underutilized in the analysis of critical care outcomes. The availability of high-quality data from a large number of well-characterized critically ill patients across several studies provides a unique opportunity for these investigations. The proposed studies may help to improve the way we conduct clinical trials and how we evaluate the effects of interventions on the occurrence of important morbidity outcomes. Specifically, the use of analytical approaches that account for the effects of time may improve our ability to evaluate treatments more efficiently with studies of shorter duration and fewer patients, evaluate benefit or harm earlier, detect interventions that may ultimately be beneficial as a component of patient care despite no effect on mortality, or aid clinical decision making in real-time. Specific aims are:
Aim 1: Characterize the effects of interventions on morbidity outcomes through time in critical care trials using a novel analytical approach.
Aim 2: Describe the association of prognostic factors measured repeatedly through time and clinical outcomes in patients with acute lung injury enrolled in ARDS Network trials.
Aim 3: Collect prospective data in acute lung injury patients to describe the association between plateau pressure measured repeatedly through time and the occurrence of clinical endpoints.