Consolidated Radiology Complex: high tech in the shade of the palms
Puerto Rican radiology institute speeds up reporting by 40% using speech recognition
Initial challenge
The health reforms of 1993 led to the privatization of public facilities, with the government financing a medical insurance program contracted to the private sector. By late 2000, all municipalities had been incorporated into the health reform plan, with 99% insured and 1.8 million participants in the Government of Puerto Rico health insurance plan. Puerto Rico's current healthcare system has around 13 public and 45 private hospitals with a total of more than 12,000 beds. Specialized institutions act as service providers to hospitals and general practitioners; one of them being the Consolidated Radiology Complex serving the southern and northeastern areas of the island.
Radiology consultations in particular are critical to diagnostics and treatments and many hospitals and GPs depend on the fast delivery of the results to be able to continue treatment. Amazingly, time is often wasted dealing with administration - with long waiting times between the dictation of the report by the radiologist and the transcription of the report by the transcription team. Reasons for the transcription bottleneck vary from cost restrictions to the difficulty of finding good medical transcriptionists. In this respect, the situation in Puerto Rico is very similar to that in mainland USA.
CRC's president, Dr. Jorge Haddock, decided to make a difference by implementing speech recognition in order to solve both of the aforementioned issues: to reduce the costs associated with transcribing medical reports and increase the productivity of the transcriptionist in an attempt to facilitate faster diagnosis and medical reporting.
Identifying improvement potential
CRC had been using the digital dictation system from Canada-based Crescendo Systems Corporation. According to Marirosi Martinez, CRC's transcription manager, digital dictation has been a relief, but falls somewhat short in handling the increasing number of patients. They either had to invest in new transcription staff, accept long waiting times for reports or outsource more work to external service providers. "Outsourcing is an option, but there are quality and confidentiality issues which, in the end, prevented us from resorting to additional outsourcing services," Marirosi explains.
Another option was speech recognition, which would provide the transcriptionists with a transcribed draft of the dictation, so that they only had to correct it. In September 2005, after an extensive evaluation period, Marirosi and her team decided to add speech recognition to their digital dictation solution. Crescendo's system is powered by the SpeechMagic speech recognition technology from Philips, supporting network-based architecture, interfacing with other healthcare applications and providing a specialized recognition vocabulary (ConText) for radiology. Because the six radiologists at CRC dictate their reports in English, the US Radiology ConText has been implemented.

Communication is everything
In fact, change management is one of the key factors when it comes to implementing speech recognition. Marirosi noted that many people are resistant to change, and is convinced that open communication is the key to reducing any anxieties related to the introduction of new technologies.
"Prepare everyone who is to be involved in the process. Prepare the transcriptionist; distribute all the information you have. Talk to the doctors, administrators - tell them what they will find and what they won't find in the new system," she recommends. "And above all, define exactly what you expect from the system for now and for the future."
Crescendo took CRC's requirements into account and Marirosi is highly satisfied with the way they were implemented. "Crescendo's service is excellent," she says. The geographical distance from Puerto Rico to Canada, where Crescendo is based, is no problem, as Crescendo engineers can remotely log in and administer CRC's system. However, when the system went live, IT engineers from Crescendo were onsite to ensure a smooth transition. "No matter how good the preparation is, you will always find things that are different from what you expected. Solve them as early as you can. Don't wait; it will cause you a lot of trouble and cost you a lot of money," advises Marirosi. Thanks to the fast and professional service received, CRC now has an "excellent" speech recognition system, something on which the facility can build for the future: "We are already looking into ways of interfacing the report creation system with the Radiology Information Systems," says Marirosi. This will further speed up radiology reporting, as the patient data will automatically be included in the dictation.
CRC is on the right track and now has a reporting system that complies with international standards. Report turnaround time has been reduced. MRI and CT scans, which generate long reports, are now available within 3 to 5 minutes, compared with 5 to 10 minutes previously. This reduces the waiting times for patients and referring physicians alike, as well as the costs associated with medical transcription.
"I really benefit from the time saved through speech recognition," Marirosi expresses. "Managing the team has become easier and more effective, as stress levels have been reduced and the quality of the reports has improved."


- 28,000 lines transcribed per month
- 40% dictation volume between 2005 and 2006
- 30-40% acceleration in report turnaround since the implementation of speech recognition
- DigiScribe-XL Dictation
- Crescendo Speech Processing powered by SpeechMagic
- MedRouter Workflow Management & Distribution
- MD Center-XL Signature
