by Frank Zadravecz, MPH
Across industries, Big Data is widely believed to offer a higher form of intelligence and knowledge that can generate insights that were previously impossible. However, the current cacophony of patient alarms we experience on the wards is not as informative for medical providers as we would hope. Are these alarms adding noise to otherwise informative Big Data?
In the June 2015 issue of The Hospitalist, Karen Appold highlights the unique position that hospitalists hold in the fight against the overwhelming number of alarms generated by patient bedside monitors. In an interview with Ms. Appold, Vladmir Cadet of the ECRI Institute frames alarm load in the context of our current healthcare system, and suggests that alarm fatigue has become an enterprise-wide issue potentially due to sector-wide implementation of error-prone systems. The presence of non-specific alarm settings on our wards hints that alarm mismanagement may be rooted deeper than at the level of individual end-users, instead stemming from a combination of competing hospital priorities, failures to analyze patient alarms comprehensively, and reliance on purely reactive solutions when critical notifications do occur.
Clinician and researchers are well aware that simple aggregation of patient data does not provide a ready tool to stave off adverse events in the hospital. There are mountains of information available to us that we need to wade through and so we have developed early warning scores and patient risk stratification tools to identify the most critical information for patient care. These scores, sometimes calculated by hand and often automated, have the potential to increase our ability to effectively multitask between seemingly never-ending data streams.
Is there potential to “double-dip” and use these automated tools to improve the value of inpatient alarms?
As a 2015 Society for Hospital Medicine Student Hospitalist Scholar, I am investigating the frequency and accuracy of physiologic alarms at our institution, and measuring the impact of nursing alarm load on the safety of our ward patients. As I document the associations between alarm quality, quantity, and patient adverse events, I hope to suggest hospital-level changes in alarm management that leverage systems readily built into our current clinical workflow.
Formal alarm management is more than reducing the number of alarms sounded. We envision a viable solution to improve alarm accuracy that employs our established and automated risk stratification tools. Repurposing these systems, we could feasibly recalibrate alarm thresholds for individual patients based on their current stratum, and allow thresholds to vary over time as a patient’s clinical trajectory improves or deteriorates during their ward stay.
As in every quality improvement project, our ability to answer specific questions is limited by the quality of data we collect. We have made significant strides matching alarms data to the clinician notified, but we are struggling to map individual bedside monitors and their subsequent alarms to the patients who generated the physiologic data. Bridging these data sources that were not originally designed for this purpose has proved difficult, but with some ingenuity in triangulating clinician and patient administrative data we have a promising solution.
The potential impact of my project findings on patient safety efforts at our hospital will depend on the level of key stakeholder buy-in and institutional backing. We have purposefully designed this project to simultaneously measure actionable patient adverse events and define a baseline for our future efforts in alarm management in order to remove any substantial barriers to quality improvement following our report.
We’re looking to use our Big Data – without the headaches.
Frank Zadravecz, MPH, is an alumnus of the Columbia University Mailman School of Public Health, a rising second year medical student at the University of Illinois at Chicago College of Medicine, and a Society for Hospital Medicine Student Hospitalist Scholar at the University of Chicago Medical Center. You can follow Frank on twitter: @frankzadravecz