Workflow analysis yields process refinements

SAN ANTONIO - Workflow analysis can be a valuable approach to improving efficiency and productivity, both before and after change, according to a presentation at this week's PACS 2006: Digital Healthcare Information and Management Systems (DHIMS) conference.

"Graphical workflow analysis can be a very powerful tool for discovery, for planning, and for evaluation," said Dr. Steven Horii, who is from the University of Pennsylvania Medical Center in Philadelphia. He spoke during a session at the PACS conference, which is sponsored by the University of Rochester School of Medicine in New York.

A workflow study is useful for justification; finding hidden bottlenecks; planning for a PACS, RIS, or any other information system; planning for a new physical plant; or to re-engineer your processes, Horii said.

If you're going to do a workflow analysis, it's a good idea to perform statistical tests to ensure comparisons are valid and that differences are significant. Also, don't forget the impact of a learning curve, and avoid the common mistake of taking measurements immediately after the introduction of a new technology, Horii said.

"People take a while to get used to something new," he said.

To perform a workflow analysis, institutions need to identify the main questions to be answered, and define the scope, goals, and metrics of the analysis, Horii said. Workflow can then be observed and measured, followed by analysis of the results.

The main questions involve determining the current processes and what problems exist. Other areas include identifying what has been done, if anything, to try to remedy the problems. Institutions also need to ask what new ideas or remedies are being considered, and what are the driving forces behind the need or desire to change processes, Horii said.

As for scope and goals, this is usually a service or operation that is the subject of the workflow analysis, such as replacing film with computed radiography, adding a multidetector CT system, or changing from conventional transcription to speech recognition, he said.

While the scope of the analysis should be restricted to that service or operation, a change in one area will likely impact others. The scope will also be much larger, for example, with a change from film and paper to digital healthcare information and management systems, Horii said.

Workflow analysis metrics to measure include time for a task, the number of steps in a task, who performs each step, and what information is input and output at each step, he said. Other metrics include error or failure rate of the step, remedies for the failure, and resources required to perform the step.

The observing and measuring part of the process is labor-intensive, and often requires an observer to follow personnel around and record what they do, when they do it, and how long it takes, Horii said.

"You cannot do this without cooperation from your personnel," he said.

Also, be aware of the Hawthorne effect, in which people's behavior changes as a result of being observed, Horii said.

The analysis component of the process is usually statistical, and it's a good reason to involve a biometrician or industrial psychologist, Horii said.

In practical terms, performing workflow analysis should begin by convening the people involved in the process and asking them to describe in detail what they do, he said. A draft diagram or chart representing the workflow steps should be created, and the diagram should be modified based on critiques generated during discussion.

The final workflow diagram should be based on descriptions and comments, and is a graphical workflow analysis process.

"A diagram conveys a lot of information about decision points and alternate steps much more easily than a text list of process items," Horii said.

At UPMC, this type of workflow analysis has paid dividends in its intraoperative ultrasound operations, according to Horii. Because intraoperative ultrasound often seemed to cause major disruption for radiology as well as surgical delays, researchers sought to determine why the problems occurred and whether there were ways they could intervene.

From the analysis, the researchers determined that the major problem was the difference between a scheduled and nonscheduled intraoperative ultrasound, Horii said.

"By using this methodology, we have been able to convince the surgeons that letting (the radiology department) know in advance that they will need us is very likely to result in shorter delays for them and shorter delays for us," he said. "So they've been much more cooperative about this since we went through this effort."

Before the change, users can examine the workflow diagram and predict steps that will be changed or eliminated. After the change, institutions can verify that the re-engineering actually worked, or if it inadvertently created more steps, Horii said.

By Erik L. Ridley
AuntMinnie.com staff writer
March 17, 2006

Related Reading

Digital dashboard eases radiology IT operation, March 16, 2006

Russian teleradiology network meets needs, exceeds expectations, March 6, 2006

Higher monitor luminance levels may hinder performance, February 27, 2006

PACS doesn't hurt pediatric ER performance, February 20, 2006

Canadian radiologists pursue lossy compression research, February 16, 2006

Copyright © 2006 AuntMinnie.com

Page 1 of 603
Next Page