An adverse event screening tool based on routinely collected hospital-acquired diagnoses

Abstract

The aim was to develop an electronic adverse event (AE) screening tool applicable to acute care hospital episodes for patients admitted with chronic heart failure (CHF) and pneumonia. Consensus building using a modified Delphi method and descriptive analysis of hospital discharge data. Consultant physicians in general medicine (n = 38). In-hospital acquired (C-prefix) diagnoses associated with CHF and pneumonia admissions to 230 hospitals in Victoria, Australia, were extracted from the Victorian Admitted Episodes Data Set between July 2004 and June 2007. A 9-point rating scale was used to prioritize diagnoses acquired during hospitalization (routinely coded as a 'C-prefix' diagnosis to distinguish from diagnoses present on admission) for inclusion within an AE screening tool. Diagnoses rated a group median score between 7 and 9 by the physician panel were included. Selection of C-prefix diagnoses with a group median rating of 7-9 in a screening tool, and the level of physician agreement, as assessed using the Interpercentile Range Adjusted for Symmetry. Of 697 initial C-prefix diagnoses, there were high levels of agreement to include 113 (16.2%) in the AE screening tool. Using these selected diagnoses, a potential AE was flagged in 14% of all admissions for the two index conditions. Intra-rater reliability for each clinician ranged from kappa 0.482 to 1.0. A high level of physician agreement was obtained in selecting in-hospital diagnoses for inclusion in an AE screening tool based on routinely collected data. These results support further tool validation.