In an era of declining reimbursement for radiology, one practice is testing the limits of IT’s ability to improve productivity In retrospect, 10 years ago, it took us many months to accomplish our first DICOM modality integration of a RIS and a CT scanner at Inland Imaging, Spokane, Wash. The standards were vague and communications between systems were difficult and incomplete, but at that time, it was a great accomplishment. Although other industries had done this form of machine-to-IT integration for some time, it was relatively new for health care. Today, we have more than 500 different units (representing modalities that include CT, MRI, ultrasound, and digitizers) connected via DICOM to our RIS/PACS, and we can add a new device in a matter of minutes. The ability to share health information via DICOM, HL7, and other standards has come a long way in the past decade. At Inland Imaging, our imaging centers achieved efficiencies using IT (RIS, PACS, and voice recognition) that enabled them to handle increasing volumes with the same, or significantly fewer, medical records personnel, drivers, transcriptionists, and core (front-desk) staff. Our technologists display their patient worklist via DICOM Modality Worklist for the day and take control of that workflow. We no longer print and hang film. We no longer have a need for darkrooms. We do not have to chase down prior films or worry about them being delivered in time to view along with the current examination because most priors are already on our PACS. We no longer hand deliver more than 20,000 film jackets monthly. Our referring physicians and other health care providers access reports and images online. We have more than 5,000 remote user accounts on our Web-based report- and image-viewing system today. Our radiologists perform remote consultations with referring physicians while both are looking at exactly the same images, without either physician having to leave his or her office. Our technologists and radiologists have the same real-time communication and consultation capability. Our report-turnaround times are dramatically improved due to voice recognition. Patient care has also improved; we have made great progress. William Keyes, MD, president of the Inland Imaging division, is an Inland Imaging neuroradiologist and is in charge of IT. He says, “PACS and DICOM Modality Worklist technology help our technologists by providing a greater degree of standardization and quality control. Our standardized PACS presentation states facilitate the training of the technologists and allow them to rotate easily through different areas.” Key to Subspecialization The success of PACS and related technologies in moving exams to the radiologist—versus moving the radiologists to the exams—has allowed a growth opportunity in subspecialization. Radiology groups have grown much larger to generate the volumes that enable them to subspecialize. These larger groups tend to serve multiple medical centers and hospital systems, multispecialty clinics, and other nonradiologist-owned technical imaging service providers. This fragmentation of workflow is particularly challenging when coupled with expectations of higher productivity and higher quality. As Inland Imaging expanded to different cities, serving multiple health care entities, a whole new set of efficiency and quality problems emerged. We encountered many disparate information systems. The level of the different IT staffs’ RIS knowledge and ability, and their willingness to cooperate in IT strategies, varied dramatically. We were expected to use multiple PACS viewers and dictation systems, and to sign reports in multiple systems. We were receiving exam information from five or more RIS types, three kinds of hospital information system (HIS), three voice-dictation or voice-recognition systems, and other systems. Not only were there technical hurdles, but security and general data-sharing policies also needed to be aligned.

Our mantra here was kill the parrot, meaning reduce dictation of data elements that can be captured upstream elsewhere in the process and automatically inserted into the report.

In order to move examinations around seamlessly, we decided to provide our own common IT infrastructure for workflow, productivity, and quality. We called this our franchise model of IT. We also needed to convince our customers that our own franchise IT model was of as much benefit to the hospital or other entity as it was to Inland Imaging, and that it would not disrupt their existing IT solutions. We chose to build a layer of software technology to rest on top of the assorted RIS, PACS, and other health IT systems that we encountered. Our goal became having one radiologist worklist, one voice-dictation/voice-recognition system, one report-signature queue, and one aggregated source of data to use in analyzing productivity, quality, and a number of other factors, all independent of the host or source systems. Using interface engines and application programming interface tools, we were able to make our own software layer relatively transparent to the host or originating PACS, RIS, HIS, electronic medical record, or other system. In the design of our new workflow systems, we avoided causing problems for the existing systems already established in the assorted clinics and hospitals that we served. We built systems to fill in the gaps that we saw in meeting our needs. We built dispatching, worklist-monitoring, and electronic forms systems so that we could move exams to Phoenix, Seattle, or Spokane—wherever the radiologist best suited to interpreting that examination was situated—without faxing paperwork. Maintaining Quality Increasing the overall velocity of radiology exams through the enterprise required changes in quality-control systems. The radiologists responded by establishing a physician-led quality committee and a small department focused on quality control and measurement. Today, that department focuses on:

• standardizing report formats and general content by subspecialty or division;
• reviewing a statistically significant sample of physician reports for errors, which are often related to voice recognition, and giving findings back to each dictating radiologist and to the IT staff to fine-tune the voice-recognition systems;
• assisting in performing structured, formal peer review on a routine basis; and
• establishing and supporting critical- and urgent-findings tracking systems and processes.

In describing the resulting Inland Imaging quality initiative, Keyes says, “We measure quality through our interactive peer-review system. The peer-review system needs to be simple and readily accessible; otherwise, it is not going to be used. Our focus is on giving the radiologist more exam-viewing time by requiring less time to be spent on activities like dictation, exam sorting, and clinical and procedural information gathering.” He continues, “We have created individualized and exclusive worklists to prevent lost studies, and we electronically tally the work each day in order to avoid wide fluctuations in workload between similar work slots. In addition, we can begin viewing the study while it is in progress. We no longer need to wait for all the paperwork because it is all immediately available online. This is of particular importance with stroke and other emergency-department–related imaging.” Our goal was to increase productivity and quality while eliminating as many hassle factors as possible. As we proceed in our efforts to improve throughput and quality with systems and process improvement, success is not always linear. Our systems, in some cases (at least initially), have actually slowed us down or shifted work, causing some disruption. For example, although our report turnaround has greatly improved due to voice recognition, and the radiologist no longer batch signs reports 24 hours later, while trying to remember the nature of the exam, voice recognition has placed the burden on the radiologists of correcting some errors that were previously corrected by transcriptionists. To remedy this, we created standardized report templates, by specialty, to improve report consistency while reducing the rare misses of the voice-recognition systems. Kill the Parrot Another example is that we now have the ability to insert signs, symptoms, and other data into our reports automatically, thus reducing dictation time for the radiologist. Our mantra here was kill the parrot, meaning reduce dictation of data elements that can be captured upstream elsewhere in the process and automatically inserted into the report. As a consequence, we are working on getting the upstream data collected more accurately for these inserted data elements. The spelling and word case need to be perfect when they are entered upstream or the radiologist will need to make corrections to the inserted data, and might as well dictate them. We are making a process-improvement effort to identify the workflow and the key points at which these data need to be validated. This level of accuracy was not previously expected of our technologists, assistants, and schedulers, and achieving it may take more time on their part, but this will ultimately save the radiologist dictation time. Another challenge that we continue to address, as a group, is that one technology solution or workflow does not fit all modalities or workflow types. We continue to experiment with applying different configurations of our systems to create faster workflow scenarios while maintaining quality in screening mammography, radiography, and other areas with extremely high volumes. Many people have asked me when our systems will stop evolving. The answer is the day we stop doing radiology. We are all under pressure to increase productivity and dependence on IT. We believe that process improvement and quality assurance are crucial to leveraging IT. We realize that getting things right the first (or even second) time around may not happen; that’s why they call it continuous process/quality improvement. As long as we are willing to make efforts to improve quality and productivity together, our push for productivity will be successful.