View Abstract

A5084 - Contributors to ICU Clinicians' Predictions of Patient Outcomes: A Qualitative Analysis
Author Block: A. Buehler1, I. Ciuffetelli2, A. Delman3, S. Kent4, D. Bayard5, E. Cooney5, S. D. Halpern6, M. E. Detsky7; 1University of California San Diego School of Medicine, La Jolla, CA, United States, 2McGovern Medical School at University of Texas, Houston, TX, United States, 3Wayne State University School of Medicine, Detroit, MI, United States, 4University of Kentucky, Lexington, KY, United States, 5University of Pennsylvania, Philadelphia, PA, United States, 6Univ of Pennsylvania, Philadelphia, PA, United States, 7Medicine, Sinai Health System, Toronto, ON, Canada.
Rationale: The prognosis of critically ill patients is fundamental to shared decision making. How physicians and nurses generate their prognoses is not well understood. The objective of this study was to explore which patient factors physicians and nurses consider when making prognostic assessments for critically ill patients.
Methods: Physicians and nurses were asked to predict 6-month patient outcomes and to describe, in an open-ended response, what patient factors they considered in making each prognosis. Principles of grounded theory were used to analyze the free-text statements.
Results: Among 303 patients, 542 surveys were completed (254 for physicians, 288 for nurses). These surveys included predictions of 6-month mortality as well as an open-ended response explaining the factors considered in making the prognoses. We identified 23 contributing factors within 8 major categories. Overall, the three most frequently reported factors for both physicians and nurses were: acute conditions (117/254, 46% of physician statements; 145/288, 50% of nurse statements), medical history and co-morbidities (68/254, 27% of physician statements; 82/288, 28% of nurse statements), and trajectory (50/254, 20% of physician statements; 64/288, 22% of nurse statements). For predictions of six-month survival, physicians were more likely to mention procedures (25/32, 78% of statements reporting this factor) and age (25/33, 76%), while nurses were more likely to mention prior experience (25/29, 86%), current functional status (29/36, 81%), social support (25/29, 86%) and behavioral factors (33/38, 87%). For predictions of six-month mortality physicians (30/39, 77%) and nurses (20/31, 64%) were more likely to mention cancer. For physician statements, the factors reported with the highest proportion of correct predictions of six-month survival/mortality
included definitive treatment (20/23, 87% of statements reporting this factor), procedures (27/31, 87%), age (24/30, 80%) and acute conditions (92/117, 79%). For nurse statements, the factors reported with the highest proportion of correct predictions included current functional status (29/36, 81%), behavioral (30/37, 81%), procedures (27/32, 84%) and acute conditions (109/144 76%).
Conclusions: ICU physicians and nurses report using a variety of factors to inform their prognostic estimates, with the acute illness, underlying comorbidities, and recent trajectory being the most common contributors for both types of clinicians. The most commonly reported factors were not necessarily associated with a higher degree of predictive accuracy, while some infrequently reported factors were associated with a high predictive accuracy.