To determine whether a cluster of infections in a hospital represents an outbreak, hospital infection control professionals must first recognize a new pattern of infection and then determine whether the pattern is of sufficient concern to merit intervention. Problems during the recognition and investigative processes incur delays, and with delays come increased costs to the hospital, inpatient morbidity and mortality, and in economic terms -- sometimes exceeding $1 million. Recent academic studies of outbreak alerting have focused on syndromic surveillance algorithms, which utilize pre-clinical data (e.g., records of over-the-counter pharmaceutical purchases and of chief complaints from emergency room visits) in an attempt to detect outbreaks in large geographic areas. Based on the observation that simple detection methods applied to hospital data have already shown promising results, the current project begins with the presumption that new, more advanced approaches to computer-assisted hospital infection control can potentially improve patient outcomes nationwide. This study aims to develop MIASMA, a Medical Informatics Application for Systematic Microbiological Alerts. MIASMA will use electronic medical record (EMR) derived culture data and patient-specific location and order data abstracted into an institution-independent standard format as input. It will incorporate statistical and rule-based alerting methods along with novel heuristic detection methods. MIASMA will also collect, organize, and display the data needed by hospital infection staff to investigate alerts, including a visualization component supported by geographic models of hospital locations. It will be deployed and evaluated at Vanderbilt University Hospital, but will be made freely available to any institution with the necessary data sources available.
This CDC dissertation research project describes planned construction and evaluation of MIASMA, a computerized system for alerting hospital staff regarding potential bacterial outbreaks. MIASMA will help hospital infection control staff detect and investigate outbreaks more quickly, and if successful will thus reduce further spread of outbreaks to additional patients. MIASMA will be designed flexibly using open source software, and after an evaluation to document its efficacy, we will release the MIASMA software the general public via a free software license with a goal of allowing hospital with the necessary data sources available to be able to use the system.
