Hospitals turn to big data to identify risks in the ICU

Intensive care units (ICUs), seeking to reduce avoidable deaths, can harness big data to identify trends that may increase patient risks, according to the Wall Street Journal.

ICUs admit more than 5 million patients a year, and mortality rates range from 10 percent to 29 percent. Many of these deaths have roots in preventable complications, such as hospital-acquired infections, blood clots and delirium due to oversedation. Hospitals have taken numerous steps to address avoidable ICU complications such as investing in more robust infection prevention programs. Now, however, the WSJ reports, some hospitals are testing an approach that incorporates years of medical data and records, many of which have never been analyzed together before.

While simpler strategies such as checklists have helped ICUs reduce more predictable risks, less predictable hazards are a greater concern for healthcare providers, Kenneth Sands, chief quality officer and senior vice president at Beth Israel Deaconess Medical Center, told the WSJ. Beth Israel, in collaboration with experts from the Massachusetts Institute of Technology and information technology company Aptima Inc., is developing an initiative called Risky States, which uses big data to gauge an ICU's risk level.

The project team analyzed data from patients in Beth Israel's seven ICUs from 2012 to 2014 and found multiple scenarios that correlated with increased risk--the "risky states" of the program name--including higher numbers of sicker patients, a higher percentage of nurses with less than a year of experience and high patient-to-nurse ratios.

They next identified about 30 patient harms that occurred during or after the risky states, including medication errors, readmission to the ICU, bleeding and cardiac arrest. Beth Israel and its partners developed an application that pulls this data from electronic medical records automatically and gives clinicians the option to add any further concerns. Based on this information, the app tabulates an overall risk score in real time.

Baltimore's Johns Hopkins Medicine is pursuing similar strategies through a program that compares current and historical patient data. To give all clinicians equal access to the same data and risk information, Hopkins needs better connectivity among its medical devices, Peter Pronovost, M.D., Ph.D., senior vice president for patient safety and quality, told the WSJ.

To learn more:
- here's the article