DESCRIPTION (provided by applicant): Delirium is a common, costly, life-threatening, and potentially preventable problem for older hospitalized persons. While increased mortality following delirium has been well-established, the contributions of delirium itself and its complications to mortality remain unclear. The broad objective is to elucidate the pathways to mortality associated with delirium, and to examine the relative contributions of baseline vulnerability, precipitating factors, and factors occurring after delirium onset that are related t delirium and its complications. The specific aims are: (1) to determine the independent contribution of delirium to increased mortality during hospitalization and up to 1 month follow-up, after accounting for baseline vulnerability and precipitating factors; (2) to examine the pathways to mortality associated with delirium, focusing on post-delirium factors which are potentially preventable; and (3) to evaluate the magnitude of the impact of these potentially preventable post- delirium factors on mortality. We propose secondary data analysis of an existing NIH-funded dataset of 919 consecutive hospital admissions with complete tracking of all deaths. This sample provides a large-scale dataset utilizing state-of-the-art methods, including cognitive testing and the Confusion Assessment Method, to prospectively assess delirium, the main predictor variable which occurred in 13%. Information on key study variables and co-variables came from interviews with participants, family members, and hospital nurses, and medical record review. Information on deaths (including dates and causes of death), which occurred in 7%, came from family interviews, daily obituary review, medical record review, National Death Index, Medicare and Social Security databases. Prior to analyses, an interdisciplinary Expert Panel will review all of the post- delirium factors and causes of death to rate each on whether they can be plausibly attributed to delirium (low- medium-high likelihood). We will utilize appropriate multivariable survival analysis techniques to examine the relationship of delirium and mortality while controlling for important co-variables. Separate propensity score models will be constructed for baseline vulnerability and precipitating factors to examine their relative contributions to delirium and to death. We will analyze post-delirium factors utilizing appropriate bivariate then multivariable approaches to identify the final set of mediators. The identified mediators will be included in a final model to determine their joint effect. Pilot work has confirmed adequate power for our proposed aims. Strengths of this high-risk/high-gain study include the clinical relevance of the topic; innovative, cost- effective use of an existing clinicaly-rich dataset with extensive information on baseline, precipitating, and mediating factors; state-of-the-art delirium assessment with precise time of onset; complete mortality tracking with causes of death; and the expertise of the research team. This project holds substantial promise to produce new knowledge that will help to identify potentially preventable factors on the pathway between delirium and death, which will form the basis for future intervention strategies to address this important area.
