Computers Lead to Safer Blood Transfusions, Chemotherapy

Newswise — Computer scientists at the University of Massachusetts Amherst are working with Massachusetts health care professionals to analyze medical procedures, including blood transfusions and chemotherapy treatments, with the goal of improving patient safety. The team is also analyzing the flow of patients in emergency rooms to reduce waiting time.

"Health care workers are dealing with new machinery and medical activities that are increasingly complex, and a 1999 report by the Institute of Medicine found that medical accidents account for almost 100,000 deaths in the U.S. each year," says Lori Clarke, a professor of computer science. "Computers can help by detecting flaws in the processes used to deliver medical care, and confirm that efforts to fix the flaws don't create other problems down the line." Additional researchers from UMass Amherst include Leon Osterweil, a professor of computer science; George Avrunin, a professor of mathematics and statistics; Elizabeth Henneman, a professor of nursing, and computer science graduate students Bin Chen, Stefan Christov, Rachel Cobleigh, Huong Phan and M.S. Raunak. Phillip Henneman, former director of emergency services at the Tufts-Baystate Medical Center in Springfield and Wilson Mertens, medical director of cancer services at the D'Amour Center for Cancer Care in Springfield, are also part of the team.

"Eventually, we hope that our technology will be used by the Joint Commission that accredits and certifies health care organizations, and by medical informatics and statistics personnel based in hospitals," says Clarke. "This technology could also be applied to medical processes performed by pharmacists, nursing home workers and home health care workers." One of the first procedures the team analyzed was a process for performing blood transfusions, which is based on a national standard, and is representative of blood transfusion processes being used at hospitals throughout the country. The team chose this procedure because adverse events, including giving patients the wrong type of blood or giving blood to the wrong patient, have been reported nationally, and have caused serious harm and even death.

The objective was to isolate flaws in the process that might lead to these adverse events, so these flaws could be addressed by health care professionals in order to make blood transfusions safer.

One flaw the analysis revealed was a "deadlock," something that software engineering techniques can detect. The deadlock, essentially a situation where the participants would have to wait endlessly, occurs in cases where a nurse submits a request for blood to the blood bank, but the blood bank needs the nurse to determine the patient's blood type first. In this case, the nurse and the blood bank both wait for each other, causing a delay that could create serious problems for the patient.

The medical professionals were aware that such situations had arisen, but were less sure about why they occurred and how to prevent this situation. Technologies developed by the team indicated the specific cause of the problem, and verified that a proposed modification to the process, namely requiring nurses to check for the availability of the blood type before notifying the blood bank, did indeed prevent its recurrence.

Delivery of chemotherapy and radiation used to treat cancer patients was also analyzed. "Dosage for chemotherapy depends on body surface area, an important calculation based on the patient's height and weight," says Clarke. "We found that there were ways of going through this process in which height and weight were measured only once, often at the beginning of treatment."

"This created the dangerous possibility that significant changes in weight, which are quite common in the course of chemotherapy, might go unrecognized," says Osterweil. "This could potentially lead to incorrect dosages of chemotherapy medications."

The team is also starting to build simulators for modeling the flow of patients and the use of resources in emergency rooms, which will hopefully decrease the long waits endured by many emergency room patients.

To analyze the medical processes, researchers employed software engineering tools usually used to define and analyze complex software systems, including a special language called Little-JIL that had been developed by Osterweil and his team. The project will eventually result in a suite of tools that will be made available as open source prototypes to be used by other researchers.