A group of researchers from Massachusetts General Hospital cut the incidence of delayed image delivery to PACS by two-thirds by using an automated system that alerted managers when the time-to-PACS (TTP) had exceeded a certain threshold. The internally developed web application—called Tempus Fugit—demonstrates the potency of targeted workflow intervention in an imaging department.
Their research was published in the Journal of the American College of Radiology.
Modern-day radiology departments are a production line. Exams must pass through a series of steps that depend on the completion of the previous one: scheduling the patient, performing the exam, sending images from the scanner to PACS, reading the exam, dictating the interpretation, etc.
This causes delays to have an outsize influence, picking up intensity as they radiate out through the system—a major problem at a large hospital like Massachusetts General, which preforms more than 90,000 imaging examinations every month.
“To solve this problem, we developed an application named Tempus Fugit (TF),” wrote lead author Oleg Pianykh, PhD, et al. Pianykh is an Assistant Professor of Radiology and Director of Medical Analytics at Massachusetts General Hospital, Harvard Medical School.
“TF tracks examinations, identifies outliers, and alerts radiology supervisors to take immediate actions. In this work, we studied the effect TF had on one of the most critical workflow metrics: time-to-PACS (TTP), defined as the amount of time it takes a technologist to send a completed examination from an acquisition device to the PACS digital archive,” the authors wrote.
TF combs through as many as 800,000 Radiology Information System (RIS) and PACS records to determine if an exam should be flagged. Pianykh and colleagues set the TTP threshold at 40 minutes, the 90th percentile of their exam-to-transmission times. They set a relatively high threshold to prevent alert fatigue or a “crying wolf” effect, according to the article.
The initial implementation featured passive alerts: managers had to watch a webpage to see if any outliers were identified. While they saw a mild reduction in the incidence of exams with delayed TTP, they saw much better results after about six months when they switched to an active alert system.
“In contrast to the incremental change brought by the TF website, the improvement after the introduction of the TF alert system was virtually immediate: the percentage of examinations with delayed TTP abruptly fell from an average of 5.8 percent to 2.3 percent,” wrote Pianykh et al. “As expected, active alerts were more effective not only because they were sent in real time, but also because they ensured that nothing could be missed, compared with time-consuming website monitoring,”
Surprisingly, implementation of TF reduced TTP for all time intervals, not just ones longer than 40 minutes—the mere existence of the system reinforced the importance of timeliness, according to the authors.
“Delays had become unacceptable in general, and not just when identified by the system,” wrote Pianykh et al. “Consequently, the 90th TTP percentile has improved from 42 minutes before alerting to only 27 minutes after the active alerts. Alerting, and mainly active alerting, has changed the entire culture of timely examination processing.”
While spikes in delays persisted around shift changes and noon—attributed to radiologists going to lunch—the magnitude of the spikes was greatly reduced.
“In essence, our computer-automated alerting was used to compensate for the intrinsic human inability to stay precisely on time,” wrote Pianykh et al. “By reminding more forgetful technologists about their work priorities, and by instilling the “time matters” culture, the TF website and alerts were able to significantly reduce TTP delays.”