CHICAGO — “Should doctors have patient data at their fingertips? The answer is an obvious yes but somehow, in radiology, we don’t practice that way,” Cree M. Gaskin, MD, of UVA said at RSNA 2014.
Radiologists experience vague reasoning and incomplete histories on a daily basis, but Gaskin argued this isn’t how it needs to be. Changing the behavior of others (read: referring physicians) can, and has been, a difficult undertaking, but what if radiologists can get the clinical information they need without digging through a system or making a bunch of calls?
“Clinical information provides context for image interpretation, it may lead to a second focused look, and you’re going to have fewer misses,” Gaskin said. It can also help radiologists provide more useful reports to their referring physicians. “Instead of a long list of differential diagnoses, we can say given the clinical context, this is most likely X.”
Gaskin cited the June 2014 medical malpractice case in which a radiologist at Brigham and Women’s Hospital missed evidence of lung cancer in a Boston woman’s chest X-ray. Johnette Ellis visited the Brigham emergency room originally in October 2006 with complaints of a persistent cough, a chest X-ray was ordered to rule out pneumonia, which was read and determined to be normal by Dr. Peter Clarke. Thirteen months later, Ellis visited the hospital again after her symptoms worsened. The CT scan showed advanced lung cancer; within seven months, the cancer spread to her kidney, liver, spine, and pubic bone before she died at the age of 47. Ellis’s daughter, who sued Clarke on her mother’s behalf, was awarded $16.7 million.
Clarke’s lawyer claimed that Clarke was not provided with Ellis’s full medical history, which included a 30 pack/year smoking history and history of lung cancer in the family.
“We can all miss things on any day, but when you have a history like that, you tend to give things a second look,” Gaskin said.
Gaskin also referenced a University of Chicago study that looked at the quality of clinical history in radiology orders, specifically with chronic conditions like Lupus or Crohn’s disease. The study identified that 40% of the time, relevant chronic conditions that would affect interpretation were not in the history provided by the referring provider. Radiologists took it upon themselves to look up the clinical history and add it to the report 35% of the time, according to the study. And 25% of the time, there was no evidence that the radiologist was aware of a relevant condition.
“The problem is related to the way most of us do things,” Gaskin said. “If you think of our classic PACS-driven workflow, you have one vendor’s health record, an order is placed in the health record and gets passed down to the RIS, which is with a second vendor, and then it is passed down to PACS, and that’s where the radiologist sits, downstream of this workflow.”
The health record has gained popularity, largely due to the Meaningful Use requirement that providers use an electronic health record. Radiologists, Gaskin said, fought this and argued that it wasn’t relevant to radiologists, who were so far downstream of the health record.
“It’s lead to a swell of patient data that sits in the EHR, but when the order is made in the EHR, as it goes downstream, you clip off almost all of the patient data,” Gaskin said. “The tech does their exam and adds the exam data, which passes downstream to the PACS, where we sit with just a little bit of order data and a little bit of exam data.”
The workflow as it stands needs to be changed, Gaskin said. There is a huge pie of patient data, but in order for it to be relevant to radiologists, it needs to be delivered to radiologists in a useful and timely manner.
While 35% of the radiologists in the University of Chicago study took it upon themselves to locate the clinical history, most radiologists won’t go through this effort.
“We read a chest X-ray in 10-15 seconds, we don’t have time to dig in a chart. We need boiled down information that we can find in 10 seconds. It needs to be fast, it needs to be filtered, and it needs to be tailored to what we are doing. The chest radiologist cares about some things that the bone radiologist does not.”
Gaskin’s institution, the University of Virginia, recognized this flawed workflow and adopted a process that most radiologists would covet.
“We said, ‘we want to access this pie [of patient data]’, and one good way of doing this is having this single RIS-health record vendor. It puts the thought of the interaction between the radiologist and the health record, and that drives the PACS. So we don’t use a PACS worklist, we use a health record worklist, and the RIS is embedded in the health record,” he said.
The lists need to be clean, easy to use, and have good prioritization, Gaskin said. In addition, the workflow at UVA includes atlases that help the radiologist compare studies to a “normal” study.
Gaskin addressed the concern that looking at more patient data, which radiologists usually don’t have access to, might slow down the process. “When we went PACS-drive to RIS-driven, one year apart, in the same six-month period, with the same number of radiologists, our exam volume actually went up,” he said. “Our highest level of priority had an 85%–86% improvement in turnaround time.”
“We’re more informed, but we read faster, probably because we are more efficient. We are not wasting time looking stuff up. We are getting reports out. No one is calling us and interrupting our reading to find out who is reading the case,” he said.
The need for clinical data isn’t a divisive topic. But Gaskin argued that clinical data is only meaningful if it’s easy.
“Clinicians are starting to coordinate care in the health record, they’re not necessarily doing it verbally or via e-mail anymore, we should be engaged with that,” he said.