Using location tracking data to assess efficiency in established clinical workflows

Using Location Tracking Data to Assess Efficiency in Established Clinical Workflows Mark Meyera MD MPH, Pamela Fairbrotherb BA, Marie Eganc RN MS, Henry Chueha MD, & Warren S. Sandbergd MD PhD a

Laboratory of Computer Science, Massachusetts General Hospital, Boston, MA b Research Assistant, Massachusetts General Hospital, Boston, MA c Department of Nursing, Massachusetts General Hospital, Boston, MA d Department of Anesthesia and Critical Care, Massachusetts General Hospital, Boston, MA Abstract: Location tracking systems are becoming more prevalent in clinical settings yet applications still are not common. We have designed a system to aid in the assessment of clinical workflow efficiency. Location data is captured from active RFID tags and processed into usable data. These data are stored and presented visually with trending capability over time. The system allows quick assessments of the impact of process changes on workflow, and isolates areas for improvement. The Massachusetts General Hospital has installed a dual emitting (active RFID & IR, Radianse, Lawrence, MA) location tracking system in its operating rooms. In this application, we utilize tracking system data to extract pertinent timestamps for discrete clinical steps in an established workflow and trend this information over time. This allows multiple ‘live’ views of process information and quick assessments of the quantitative impact of process improvement measures. We have selected needle localization of breast lesions prior to breast biopsy as the clinical workflow in which to apply location tracking data to assess and improve efficiency. Needle localization patients travel from surgery check-in to radiology to the operating room, impacting the workflow in each area. Efficient care delivery requires tight communication and coordination between units. All needle location patients are currently tracked, but location-finding requires a manual lookup and no process status can be provided. An algorithm written in Ruby cleans the data and extracts pertinent timestamps for use in process efficiency analyses. The raw location data provides the position of the RFID/IR tags in physical space at a given time, indicating physical movement of the tag and presumably the patient, but does not ensure any data validation beyond the designation that a tag was assigned to a patient scheduled for needle localization. Our algorithm performs several functions. It provides for data cleaning, removing duplicate

instances when a single receiver continually broadcasts the presence of a tag over time. It also maps receiver groups into gross clinical areas to minimize the impact of tags ‘bouncing’ between receivers and transient tag ‘drops’ from the system. Bouncing occurs when a tag is approximately equidistant in signal strength from two receivers, creating conditions where the tag transiently appears on one and then the other. Tag dropping may occur when the tag leaves the coverage area or enters an area of increased shielding where additional receivers inside the shielded area are not present. Grouping receivers by gross clinical area provides the basis for data validation. We compare actual patient location data against the expected data feed (sequence of locations) for the established clinical workflow and provide a score that represents whether the normal clinical workflow adequately fits the patient and whether the patient data should be included in the analysis. Process error detection is also enabled. Data analysis and presentation via a Rails application framework provides an online mechanism to access location tracking analysis data. We provide summary data on needle localization patients with pertinent timestamps including elapsed time in the surgery unit prior to departure to radiology, elapsed time in radiology, and elapsed time in the surgery unit after radiology but prior to the operating room. Data are presented on the patient level and averaged by day, week, and month with results graphed over time. These results are then used to assess efficiency in the needle localization process and to assess the impact of current process improvement measures. The utilization of location tracking data provides for a quantitative measure to assess process improvement in the healthcare setting. By investigating dwell times in clinical areas, we can identify variances in process flow and help address inefficiencies. In the future, we plan to extend this process to other clinical workflows and investigate additional processes in the healthcare environment.

Supported by NIH 1-R43-RR018076-01

AMIA 2006 Symposium Proceedings Page - 1031