As EMRs become more prevalent, there are greater opportunities for hospital IT leaders to leverage valuable medical data to inform future care and improve outcomes. However, this deluge of data is meaningless unless it can be organized into actionable insights.
Looking for value from raw "big data" is like looking for a needle in a haystack, according to David Goldsteen, MD, chairman and CEO of Minneapolis-based VigiLanz.
Hospitals and health systems are increasingly seeking real-time predictive analytics — which uses aggregated historical data and current patients' risk profiles to predict which patient populations have the greatest likelihood for poor outcomes — as they navigate the shift to value-based care. Because value-based care links reimbursement to quality outcomes, tools that help hospitals determine where to target resources to prevent complications and readmissions and reduce length of stay are valuable.
One form of analytics pioneered by VigiLanz is Temporalytics, which helps clinicians identify the optimal time to deliver care by analyzing the outcomes of varied response times to clinical issues. Temporalytics uses advanced statistical models and machine learning to analyze data related to the timing of clinicians' actions and compare it to the response targets set in the VigiLanz Business Intelligence system. The real-time analysis predicts scenarios where a change in the optimal response time or reduction in variability around an existing best practice can improve patient care and ultimately bolster the hospital's bottom line.
"There is a huge amount of variation in response time to whatever a triggered issue may be," Dr. Goldsteen said at Becker's Hospital Review's 2nd Annual CIO/HIT + Revenue Cycle Conference in Chicago. With Temporalytics, "the goal is to narrow variability for responses — both clinical and operational — and get clinicians to respond pretty similarly," he said.
Sign up for our FREE E-Weekly for more coverage like this sent to your inbox!
With the Temporalytics system, clinicians and administrators can identify patterns and behaviors — such as site location, day of the week or number of care personnel on staff — that contribute to slower response times while highlighting cause and effect between outcomes and cost-effective care.
Dr. Goldsteen gave the example of organism-antibiotic mismatch rule analysis. In the Temporalytics system, the rule is triggered once an organism-mismatch is discovered for a patient. The rule engine monitors activity to ensure the patient is put on an antibiotic to which the organism had been tested susceptible, according to Dr. Goldsteen. An alert will be delivered after 24 hours have elapsed following the rule activation if an appropriate antibiotic has not been ordered.
Another valuable application of VigiLanz' real-time predictive modeling system is in the fight against sepsis, which is often deadly for patients and a drain on hospitals' bottom line. In 2009, sepsis was the sixth most common principal diagnosis for hospitalization in the U.S., accounting for 836,000 stays or 2.1 percent of all hospitalizations, according to the most recent national discharge data reported by the Agency for Healthcare Research and Quality.
"It has been repeatedly proven that the earlier the intervention and treatment [for sepsis], the better the clinical and financial outcomes," said Dr. Goldsteen.
There is a clear need to improve processes that identify patients at risk for developing sepsis and provide the earliest intervention possible. However, the general model for identifying sepsis — called the systemic inflammatory response syndrome — only indicates the onset of severe sepsis after the patient has already developed it. This is clearly not an effective means of prevention and does not position clinicians to intervene as effectively as possible, Dr. Goldsteen said.
The goal is to accurately predict which patients are most susceptible to developing sepsis before it becomes severe to enable early intervention. The Temporalytics predictive model is run against every patient that comes into the hospital in real time. With this predictive model, there is much higher sensitivity and specificity in identifying the onset of sepsis at an earlier stage than the SIRS criteria. Armed with this insight, hospitals can save resources and prevent unnecessary treatment by targeting only those patients who are on the track to develop sepsis.