real-time data

What Real-Time Data Could Have Done for These Patients

by John Zaleski, Ph.D., CAP, CPHIMS
Chief Analytics Officer

The concepts behind smart alarms and use of real-time data to make better decisions are not new. Many people have had their experiences in the field of medicine and medical informatics but not all people and organizations have acted upon these experiences. Bernoulli is one organization that has.

My own experiences with conceptualizing smart monitoring and clinical surveillance go back more than 20 years to my days in school at PENN and the hospital. As part of my doctoral research and education, I was interested in identifying patients at risk for respiratory failure or decompensation. In discussion with my graduate and medical advisors, a consensus on a key clinical problem in those days was identifying patients who were viable to wean from postoperative mechanical ventilation. The opportunity to demonstrate a systems model for identifying weaning readiness involving multivariate data that was available from bedside monitoring would hold promise as an inexpensive and non-invasive method to manage the mechanical ventilation of patients who had undergone coronary artery bypass grafting, or CABG. Charting of patient vital signs was done in those days using tri-fold paper chart, and automated capture of data required developing methods and collecting data from patient bedsides using portable computers brought into patient rooms in surgical intensive care. Once cardiorespiratory data were collected, the process of determining interrelationships among data elements was performed to identify correlated behaviors that heralded predictable and informative clinical cascades pertaining to these patients.

Three specific instances pop into my head that are direct antecedents to formalizing my methodology surrounding predictive weaning behavior, and guide my vision towards clinical surveillance today. They involved three patients who experienced negative events that could have been identified and caught early enough to make different patient care plans. All three patients survived but the added complications – increased lengths of stay and added staff burden – translated into increased costs and negative impacts on the patients in the form of added trauma.

The first instance involved an elderly female patient who had received quadruple bypass surgery. She returned from the OR to the ICU and was intubated and mechanically ventilated. While she was being weaned from SIMV to CPAP, the approach may have been too aggressive for her – she was thin, small in stature, and somewhat frail. Soon after she was weaned down to CPAP (within approximately 20 minutes) she began hyperventilating and her chest tube bleeding had increased. By the time a physiologic monitoring alarm had sounded – a VTACH alarm – a code blue was called and the response team took over. In her case it was learned that the weaning had caused an increased cardiovascular load that her heart – in its strained state – could not handle. Furthermore, she was not yet ready to breathe spontaneously as she was weakened from the neuromuscular blockade, anesthetic, and the procedure itself. She spent an additional several days in ICU before being transferred to a general care unit. Had continuous, clinical surveillance, real-time data, and multi-variable predictive algorithms (i.e. smart alarms) been available to put together the various signs of distress – such as increasing respiratory rate, increased heart rate, chest tube bleeding – AND this information been provided to the care team as a ‘shoulder tap’ indicating this patient was heading south, then the care team could have intervened before she actually crashed.

The two other cases were quite similar in nature, involving middle-aged males who also had returned from cardiac bypass and aortic valve surgery, respectively. They were being monitored per postoperative CABG protocol when nursing noticed a rapid decline in cardiac output. In both instances, a rapid response was called and the root cause was related to a reduced vasopressor administration, reduced by a medical resident on the unit. The result was a direct correlation between vasoconstriction and increased afterload associated with increased resistance to flow from the weakened heart muscle. Had a smart system been available to identify the correlated events associated with reduced vasopressor and increasing heart rate, and reduced cardiac output, these patients could have caught the attention of the attending physician or nursing much sooner. As it was, both men ended up spending additional days in the ICU.

Reality is that smart systems like this do exist but not all health systems are utilizing them. Fortunately, for those that are, they will realize better patient outcomes, better staff experiences, and lower costs.

About the Author:
john zaleski

John Zaleski,
Chief Analytics Officer

Dr. Zaleski brings more than 25 years of experience in researching and ushering to market devices and products to improve health care. He received his PhD from the University of Pennsylvania, with a dissertation that describes a novel approach for modeling and prediction of post-operative respiratory behavior in post-surgical cardiac patients. Dr. Zaleski has a particular expertise in designing, developing and implementing clinical and non-clinical point of care applications for hospital enterprises. Dr. Zaleski is the named inventor or co-inventor on seven issued patents related to medical device interoperability and has authored three seminal texts on integrating medical device data with electronic health record systems and using medical device data for real-time clinical decision making. View John’s profile on Linkedin.