Computers Detect Kidney Injuries

Embedding a decision support tool in the hospital EHR increases the ability to detect acute kidney injury and can reduce the severity, and improve survival, according to researchers at the University of Pittsburgh’s School of Medicine

Acute kidney injury common in hospitalized patients, but particularly with patients in intensive care units and older adults, refers to a sudden episode of kidney failure or damage that happens within a few hours or days. It causes a build-up of waste products in the blood that can affect other organs, including the brain, heart, and lungs.

While kidney function is monitored using simple blood tests, subtle changes can elude or delay detection of a problem. Failure to recognize and manage acute kidney injury in the early stages can lead to devastating outcomes for patients and increased costs to the healthcare system.

The results recently published in the Journal of the American Society of Nephrology, addresses acute kidney injuries as one of the most costly and deadly conditions affecting hospitalized patients. There is evidence that computers with the ability to analyze changes in renal function can alert doctors of acute kidney injury before the condition is obvious clinically.

“Acute kidney injury strikes one in eight hospitalized patients and if unchecked can lead to serious complications including the need for dialysis and even death,” reports Senior Author John Kellum MD, Professor of Critical Care Medicine, Director of the Center for Critical Care Nephrology at the Pittsburgh’s School of Medicine.

In 2013, Kellum’s team released a computer program within the EHR system across 14 UPMC hospitals. The program monitored levels of blood creatinine, a standard measure of kidney function, and then was able to analyze changes noted in the blood levels.

If the levels rose too high or fast, the program fired an alert in the patient’s EHR informing doctors that acute kidney injury could be present. It also helped determine the stage of injury based on changes from the patient’s baseline kidney function.

To determine what effect the computer program was having on physician behavior and patient outcomes, Kellum’s team analyzed records from more than half a million patients admitted to UPMC and found that patients with acute kidney injury had a small but yet sustained decrease in hospital mortality, shorter length of stay in the hospital, and a decrease of 2.7 percent in dialysis rates.

“The plan is to work with pharmacists to adjust patient medications along with machine learning experts to better predict which patients will be at greatest risk of adverse events”, said Dr. Kellum. “In the future, incorporating protein biomarkers and even genomics into the system could one day revolutionize patient not for jut acute kidney injury but for other illnesses as well.”


Share Button