box10.gif (1299 bytes)







Diagnostic imaging

London-area hospitals speed up DI reporting

LONDON, Ont. – In March 2010, Healthcare Materials Management Services, a joint venture between London Health Sciences Centre (LHSC) and St. Joseph’s Health Care London (St. Joseph’s), awarded Lanier Healthcare Canada, a London, Ont.-based company, the contract to install SpeechQ after an extensive RFP process.

The system, which will be installed in the imaging departments of LHSC, St. Joseph’s and surrounding hospitals, is intended to shorten the length of time needed for radiologists’ reports to reach referring physicians, ultimately speeding up treatment decisions and improving patient care.

Previously, LHSC radiologists used a non-speech recognition transcription system to dictate interpretations. Their recorded voices would go to transcriptionists, who would transcribe the verbal reports into text. The radiologists would then review and correct the reports, and send them on to the referring physicians. The process took from 24 to 72 hours.

Using a speech recognition engine and the SpeechQ software, the radiologists’ verbal interpretation is converted to a document on their own computers, where it can be reviewed, corrected, and provided immediately to the referring physician. The report becomes available through the hospital’s health information system, as well as for distribution to external clinicians.

The project, which involved an investment of $1.4 million, has been implemented not only at LHSC and St. Joseph’s, but at 14 affiliated hospitals in southwestern Ontario, serving 107 radiologists. The technology is expected to pay for itself in 24 months through the reduction in spending on transcription services.

Implementation was a collaborative initiative between ITS team at LHSC and St. Joseph’s, and the participating Imaging departments.

Medquist’s SpeechQ is a speech recognition system that is based on a radiology dictionary. It recognizes radiology phrases quite well, and with proper editing techniques, SpeechQ learns those words that are not in its dictionary. With each report that is edited properly, the physicians’ recognition improves. While this process takes longer for some users, the majority are now experiencing excellent results.

LHSC and St. Joseph’s originally aimed to have 80 percent of its reports signed by the radiologist immediately, by-passing the transcription step, within three months of SpeechQ’s implementation, but it has exceeded its own target. The system learns over time, and now 90 percent of all diagnostic reports completed by the radiologists using speech recognition are immediately available to referring clinicians.

The remaining 10 percent of reports are sent to transcriptionists who now edit the reports, rather than type them fully, as these are already transcribed by the speech recognition software. Editing is much faster than transcribing, therefore even reports that go to an editor for corrections are turned around in less than 24 hours.

“The impact to patient care has been remarkable. At times, reports are reaching clinicians before patients even make it back to the clinic that same day,” says Kristin Sclisizzi, Clinical Lead for the Project Implementation.

“The satisfaction of clinical services has increased for other reasons as well,” comments Ali Islam, lead radiologist on the Speech Recognition Steering Committee. “SpeechQ has facilitated synoptic reporting for exams where it is appropriate.

“Structured reports for relevant exams help ensure reproducibility and interobserver consistency. For longer complex reports with multiple paragraphs, the ability to organize the content in real-time has improved the readability for the referring clinician.”

“There has been a tremendous improvement in workflow,” says David Veeneman, manager of the PACS/DI-r program for LHSC and St. Joseph’s LHSC. “Doctors are extremely pleased to have their results within a few hours of a patient’s test. They’ve reset their expectations. If they’re not seeing the results within a few hours of the test, they’re asking, ‘Why is it not available?’”

Veeneman says there were some adoption challenges, with clinicians having to learn an entirely new system. “The look and feel of the application is different, and the commands they have to use are different.”

“These are physicians who are reporting very high volumes of cases, so changing what used to be a very familiar and intuitive process for them is where the pain points are,” adds Veeneman. “The radiologists become very efficient at a process which is repeated over and over again. Trying to change habits that have been ingrained for a decade can be a challenge.”

The project was implemented on a regional basis in order to support the patient flow, and was made possible by leveraging the existing shared PACS (Picture Archiving and Communication Systems) and Diagnostic Imaging Repository. LHSC and St. Joseph’s are the primary referral sites for the surrounding hospitals in southwestern Ontario. “We have a common HIS [Health Information System] across those sites,” says Veeneman.

“The overall implementation and usage of SpeechQ has been a success,” says Dr. Islam. “Radiology interpretations, especially for acute care services, are accessible within a shorter time. Even for routine X-rays, front-line decision making is now based on a consultant radiologists’ interpretation rather than the non-radiology intern/resident.

He adds that, “With our previous system, clinical services may have waited for days before the radiology report would be generated, by which time clinical decisions would already have been made. Voice recognition technology has increased the relevancy of radiology interpretations and has ‘lifted’ the radiologist out from the basement.”

Lanier won the contract over several competitors following an RFP, presentations by the vendors and site visits. “Based on usability and quality of speech recognition, Lanier scored the highest,” says Veeneman. “It required eight weeks to install and configure the hardware and software, and six to eight weeks at each institution to implement it.

“One of the top considerations from our end-user community [in the selection process] was the vendor’s ability to support the system,” says Veeneman. “Having the support team local to the city definitely was considered a benefit of going with Lanier.”

For a London-based company like Lanier to win the London-area hospitals as customers “was considered a feather in their cap,” adds Veeneman. “It gave them the incentive to make their particular product desirable to our clinicians and helped foster a commitment to meet our needs locally.”

Lanier was the supplier of two generations of non-speech recognition systems to LHSC. The company has been marketing SpeechQ since 2006. The application is now in use by over 500 radiologists across Canada and has an average accuracy of 95 percent. (The reports become fully accurate with self-editing by the radiologists before being sent on to referring physicians.)

“Radiologists are impressed that the turnaround time is shortened so significantly that they often have the opportunity to talk with the caregiver,” says Jan Wise, CEO of Lanier Healthcare Canada. “It enhances a diagnostic imaging department’s involvement with the patient’s care.”

Early in 2012, Lanier will introduce two add-ons to the SpeechQ system. One enhancement will be additional synoptic reporting templates, which will mean reports will require less editing, be streamlined and have an even faster turnaround time.

The other enhancement will be content services that provide collaborative intelligence. This add-on is designed to reconcile inconsistencies between what a radiologist dictated and what he or she meant to dictate. The content services will also add content that may have been omitted from a report, but which the radiologist previously indicated as being important to such reports.

Posted February 23, 2012