Drilling Down
Health Data Management Magazine, April 1, 2009
When urologists at Detroit's Henry Ford Health System wanted to do a quality study around abscess drainage, they were at a loss as to where to turn for data - despite having an electronic health record at their disposal. They soon found that the homegrown EHR could not support a database search.
It's a common scenario in health care organizations, because many electronic records systems lack clinical data repositories or sophisticated search engines to sift through them. Yet, the urologists - after consulting physicians in the radiology department - were able to pull off a data mining workaround. That's because the radiology department had recently deployed a data mining tool for use in conjunction with its picture archiving and communications system.
Advertisement
"The urologists wanted to see all the procedures of robotic prostatectomy that involved abscess drainage," recalls Safwan Halabi, M.D., director of imaging informatics at Henry Ford. "That implies it was a complicated case. These physicians do a CT-guided drainage procedure. We were able to extract the complications that led to this radiology procedure."
Using the database query, Halabi was able to pull up some 30 cases that had taken place during the preceding five years, enabling a retroactive clinical analysis of what may have precipitated the drainage. Returning to the EHR, the surgeons were able to pull out the pertinent operative notes and look for patterns that might suggest ways to prevent such complications, Halabi recalls. "They can change their technique to avoid the complications," he says.
The data mining tool, from Prairie Village, Kansas-based Softek Solutions Inc., has been in place since April 2008. It works in conjunction with Henry Ford's PACS, from Philips Healthcare, Andover, Mass. The PACS has been operational for about six years and has ballooned to some 5 million reports. But the wealth of data was largely unsearchable before the deployment of the Softek tool, Halabi says. Henry Ford was so eager to get at the radiology database that it agreed to become a beta partner with Softek.
"We had several needs," Halabi recalls. "We wanted to search the database for academic purposes. If we were teaching residents the anatomy of the elbow, we could go back and look at the abnormal elbows on the fly. We also wanted to conduct retrospective analyses of disease states across our multiple imaging modalities."
Quality and safety issues were on the table as well, Halabi says. A busy imaging center with some 80 radiologists, Henry Ford performs nearly 1 million annual imaging studies. Sometimes the findings of those studies are critical, requiring immediate alerts to referring physicians. A searchable database could help flag critical cases, if not enable retrospective analysis of whether the proper communication occurred after the diagnosis was made. The Softek technology upholds these various clinical quality reporting goals, Halabi says.
Here's how the system works: First, a radiologist logs onto the Philips PACS archive. From there, the physician launches a keyword search, using the Softek search tool, which resides on the system as a companion application. Either words or phrases can be searched. Softek scans the entire PACs database, using a mirror image server that contains all the reports from the clinical application. That way, the performance of the PACS is not encumbered, Halabi notes.
Researchers are able to define what they want to see, asking, for example, for all cases involving CT exams for appendicitis patients between the ages of 10 and 20 years. The Softek search engine pulls up a list of relevant cases. "I can double-click the results returned by the query and see the images directly," Halabi says.
Halabi has used the technology to confirm the results of a dosage reduction initiative that has been under way at Henry Ford since 2005. Following the precepts of the "image gently' campaign, a national patient safety effort spearheaded by the Alliance for Radiation Safety in Pediatric Imaging (see box, above), the two-hospital health system has been promoting the use of non-ionic imaging modalities, such as MRI and ultrasound, among its younger patients, Halabi says. Using the data mining tool, Halabi was able to demonstrate that Henry Ford's educational campaign was working. Halabi analyzed data of hundreds of patients younger than age of 18 being treated for appendicitis. "We wanted to see if we were effectively reducing doses from 2005 to 2007," he says. "We found that we had successfully reduced CT doses by 25% to 30%."
Now that the technology has passed its beta test, Henry Ford will expand its use throughout its radiology department. The health system will add the search engine to its PACS workstations, letting physicians conduct their own quality studies. The organization already is creating a handful of reports, including quality reports on how well mammography results are being reported. "We can check to see if the written report has the appropriate verbiage and if the referring physicians were called," Halabi says.
Halabi would like to see some aspects of the data mining tool enhanced. For example, he would like the system to automatically feed research data to clinicians outside the radiology suite when a finding fits with a given project.
And he's not entirely bullish on data mining. While he champions the use of the technology and predicts that EHR vendors will add the capacity to their systems, he's keenly aware of the pitfalls. "You need to make sure that privacy is in place," he says. "Someone could do a search on HIV, create a spreadsheet and e-mail it out. That would cause havoc."
(c) 2009 Health Data Management and SourceMedia, Inc. All Rights Reserved.
http://www.healthdatamanagment.com/ http://www.sourcemedia.com/
For more information on related topics, visit the following channels:
