Acute lung injury (ALl) affects approximately 190,000 people a year in the United States. Despite much progress in intensive care, mortality for ALl remains high at approximately 38 percent. Many deaths occur in previously healthy people who can return to productive lives for many years if they recover. Clinical research in All is especially challenging because of heterogeneity among patients and the complexity of clinical outcomes from critical illness. This research proposal will use novel and advanced biostatistical techniques to address these challenges. We will apply these techniques to existing, high-quality data on over 2,400 patients enrolled in clinical trials of the NIH Acute Respiratory Distress Syndrome Network.
Specific Aim 1 is to develop a prognostic model specific for patients with ALI. To date, there is no classification system in use to identify patients at high risk of death. Early identification of high risk patients will have important implications in interpreting mortality statistics of new therapeutic strategies for ALI Specific Aim 2 is to use a novel analytical approach that accommodates competing outcomes, such as death versus discharge from the hospital alive, which are frequent in clinical studies of patients with ALI Analytical approaches that use logistic regression do not incorporate time when an event occurs. Important information is lost: for example, an earlier time-to-discharge is a favorable clinical outcome. Traditional models of survival analysis are likewise inadequate when there are competing outcomes. We will use an analytical approach that considers time-to-discharge and hospital mortality jointly in a statistical model. We will then apply this method to determine whether ventilator management in the 48 hours before enrollment in ARDS Network trials affected clinical outcomes.
Specific Aim 3 is to apply and validate these new analytical methods to prospectively collected data from a multicenter trial of two approaches to high frequency oscillatory ventilation in patients with acute respiratory distress syndrome. Deaths due to ALI in the United States are comparable to those from breast cancer and HIV/AIDS. Epidemiological research in ALI is especially difficult because clinical outcomes are complex and difficult to analyze. This research project will explore novel biostatistical methods to address these complexities. ? ? ? ?

National Institute of Health (NIH)
National Heart, Lung, and Blood Institute (NHLBI)
Postdoctoral Individual National Research Service Award (F32)
Project #
Application #
Study Section
Special Emphasis Panel (ZRG1-F16-T (20))
Program Officer
Colombini-Hatch, Sandra
Project Start
Project End
Budget Start
Budget End
Support Year
Fiscal Year
Total Cost
Indirect Cost
Johns Hopkins University
Internal Medicine/Medicine
Schools of Medicine
United States
Zip Code
Checkley, William; Brower, Roy G; Muñoz, Alvaro et al. (2010) Inference for mutually exclusive competing events through a mixture of generalized gamma distributions. Epidemiology 21:557-65
Checkley, William; Brower, Roy; Korpak, Anna et al. (2008) Effects of a clinical trial on mechanical ventilation practices in patients with acute lung injury. Am J Respir Crit Care Med 177:1215-22
Checkley, William; Buckley, Gillian; Gilman, Robert H et al. (2008) Multi-country analysis of the effects of diarrhoea on childhood stunting. Int J Epidemiol 37:816-30