Electronic Records Still Error-Prone, Says Study
by Brendon Nafziger
, DOTmed News Associate Editor | September 30, 2009
Even with EHR, medical
alerts are missed
Electronic records allow better tracking of information but important alerts can still slip through the cracks, according to a study published in the Archives of Internal Medicine on Monday.
Researchers at a Veterans Affairs hospital in Houston, Texas tried to figure out if the advanced electronic health records (EHR) system they had in place was eliminating a sometimes serious error: the lack of follow up after an abnormality is discovered on a test.
To do so, the researchers looked at more than 1,000 electronic "alert" notifications sent to health care providers whenever something abnormal appeared on an imaging test, such as a troubling shadow on a chest X-ray, mammogram or ultrasound.
The researchers wanted to see if the health care providers read the alerts they were sent, and then whether they took timely follow-up actions.
They found that about one-fifth of all electronic alerts were unread, but even among all those that were read, around 8 percent never received an immediate follow up.
And this temporary oversight had consequences: some of those patients whose tests generated alerts that were initially unread were later found to have clinically significant problems.
"Most test results were taken care of in a timely fashion," Hardeep Singh, M.D., M.P.H., a physician-researcher at the VA hospital in Houston and the lead author of the paper, tells DOTmed News. But surprisingly, of those that weren't followed up, "it didn't matter whether the doctor read the alert. The proportion of lack of timely follow up was about the same."
However, if a trainee health care provider ordered the tests, the alerts were slightly more likely to be unread.
VA hospitals unique testing grounds for EHRs
Dr. Singh says the study was only possible in an environment like a VA hospital, where electronic records have been in use for over a decade, and where most of the patient population seeks treatment within the system.
He also cautions that it's tricky to compare the results to those from hospitals that have yet to make the switch to EHR. "In a paper-based system, it is extremely hard to report accurately what the rates of follow up would be," he says. "You can't even track those things," such as whether a physician actually read a notice in a report. (All alerts in the electronic system record whenever they're opened to be read by the physician.)
He also mentions that a study done at Harvard in early 2000 suggested that around a third of women with abnormal mammograms never got follow ups.
"Not to compare apples to apples," he says, "but for having follow up to some type of imaging test, [with EHR] we're up to 90 plus percent."
Room for improvement
Despite its promise, Dr. Singh believes EHR still has a way to go before it reaches its full potential.
"High-tech is great," he adds, "but when we want medicine to become a high-reliability industry, we also have to think about other things."
Dr. Singh and his colleague Dean Sittig, Ph.D., in a paper published in JAMA earlier this month, call those other things the "eight rights for safe EHR use." These include the accessibility and user friendliness of the EHR software; making sure everyone follows standard medical vocabularies; properly training physicians and nurses; and ensuring that alerts don't interrupt the workflow, making them more likely to get lost.
"When you present information to doctors at certain times in their workflow," says Dr. Singh, "it makes a difference whether they would take it up or not."