Development and Evaluation of Risk Models to Predict Readmission or Death Following Discharge from an Adult General Systems Intensive Care Unit
Date
2018-07-06
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Transitions of care from intensive care unit (ICU) to ward are high-risk periods of healthcare delivery associated with ICU readmission and post-ICU mortality. Evidence-based processes for transitions are crucial for improving outcomes. Validated prediction models that include consistently associated risk factors for ICU readmission or post-ICU mortality may help to improve these practices. This mixed-methods thesis was comprised of three distinct phases: 1) systematic review and meta-analysis; 2) development of prediction models for ICU Readmission and Post-ICU Mortality using two approaches (literature-derived coefficients, data-derived coefficients [Derivation Cohort]), 3) validation of the models in an external Validation Cohort. The models for ICU Readmission showed limited discriminative ability whereas the Post-ICU Mortality models were stronger. Developing prediction models using pooled measures of association is a feasible approach, producing similar results to more the traditional data-derived method. Additional investigation to further validate the findings is required.
Description
Keywords
critical care, intensive care unit, transitions of patient care, prediction models, readmission, in-hospital mortality, meta-analysis, logistic regression
Citation
Boyd, J. M. (2018) Development and Evaluation of Risk Models to Predict Readmission or Death Following Discharge from an Adult General Systems Intensive Care Unit (Master's thesis, University of Calgary, Calgary, Canada). Retrieved from https://prism.ucalgary.ca. doi:10.11575/PRISM/32355