Abstract
Navigating through parking lots, public areas, and hallways is a stressful task for patients visiting large medical centers. Little is known about the patient experience from when they arrive at a medical center to when they check-in at their clinic. In a pilot study, we used requests for wayfinding directions from a mobile application to form a network of patient movement through the Vanderbilt University Medical Center (VUMC). From September 2016 to September 2017, patients using the wayfinding application made 3493 requests using the VUMC WalkWays application. Results show that patients frequently request directions from parking garages, on-site eateries, and the emergency room. We calculated the approximate distance patients walked to determine the extent to which associated clinical areas were co-located. Applied more generally, medical centers could use similar technologies to inform clinic placement, signage design, and resource allocation to improve the patient experience and operational efficiency.
Introduction
Navigating through the complex environment of a large medical center poses challenges to many patients1. To ease navigation, medical centers strive to design their facilities to simplify patient access to clinics, inpatient units, and procedure rooms2. Nevertheless, the task of figuring out how to get to a destination (also known as wayfinding), remains a source of anxiety for patients who visit medical centers3. Despite the best efforts in initial wayfinding design, buildings often evolve and change over time with construction projects, leading to less than intuitive layouts. Traveling throughout the medical center is a particular burden on patients with disabilities or other medical conditions that impair their ability to walk4. For certain patient populations, such as surgery or spine patients, accessibility of treatment areas from parking has been associated with patient satisfaction and perceived quality of care5. Finally, inefficient patient travel within the medical center can have an effect on hospital costs. One study at a 604 bed hospital showed that lost patients cost the organization $220,000 a year in labor costs from staff helping patients get to the correct location for care6.
Medical centers do not currently have an effective means to measure the distances that patients travel throughout the building, or to determine which areas patients frequently travel between. Figure 1 displays the continuum of the patient experience and different data sources that researchers can use to understand the patient experience. Some studies have used real-time locator systems (RTLS) to track patients in healthcare settings such as intensive care units, long term care, and the emergency department7. While these technologies are effective in promoting patient safety, increasing efficiency, and capturing important operational metrics such as wait times, they are limited to describing patient movement after they have presented at the respective unit for care. Much of the patient experience in the medical center occurs away from the clinics, inpatient units, and procedure rooms. Patient experience has also been extensively studied in home settings8,9, and in commuting to the medical center through research about access10,11. Our work seeks to fill in the gap in understanding the patient experience between when patients arrive at the medical center, and when they check-in for care. Using traditional RTLS to track patients as soon as they enter the medical center would be expensive and administratively cumbersome. Therefore, we propose that data from a wayfinding mobile application can provide insight into the patient experience within the medical center without additional data collection.
Figure 1.
Elements of the patient experience with data sources that could portray them.
Healthcare systems can use wayfinding data to infer relationships between different areas in the medical center. There is increasing interest in recent years to use network analysis of existing electronic data to make inferences about care coordination, collaboration, and social influence in the health domain12. For example, one study used the number of shared patients to form a network that reveals the connectedness of oncology specialists13. While many of these studies use data sources such as electronic health records, clinical communications, and access logs to deduce associations between entities, little work has been done to use wayfinding request data. The advantage of using wayfinding requests to infer information about patient movement is that it allows us to capture aspects of the patient experience beyond just clinical encounters. These elements of the patient experience include use of parking, restrooms, eateries, and public areas.
Healthcare organizations such as Vanderbilt University Medical Center (VUMC) would benefit from a method to passively monitor the movement of patients throughout the medical center in a cost-efficient way. The goal of this study was to use an existing data source to infer how patients move within the medical center and what areas are closely associated. The WalkWays application, developed and implemented at VUMC in November 2016, allows a patient’s phone to determine its indoor location based on Bluetooth low energy (BLE) beacons placed throughout the medical center. The interface for Walkways is shown in Figure 2. The WalkWays user can enter in a desired location and receive step-by-step directions with pictures. We believe that WalkWays request data can be used to describe the movement of patients throughout VUMC and infer networks of related medical center areas through commonly traveled routes.
Figure 2.
Screenshots from VUMC WalkWays wayfinding application14.
Methods
We collected requests for directions from the WalkWays application from September 1, 2016 to September 30, 2017. All requests in the WalkWays application are anonymous, and we did not collect any patient information in this study. WalkWays requests data was stored on the Google Analytics platform and we obtained the data through Google’s Query Explorer tool15. Each request included the starting location of the request as determined by BLE location, the desired destination, and the time the request was made. Each area in the system had an associated X-Y coordinate on a multi-building combined floor plan, where the origin was the northwest corner of the medical center. We estimated the distance traveled by the patient by calculating the perpendicular distance from patient origin to destination, that is, the sum of the difference between the X-coordinates and Y-coordinates. Next, we visualized a network of patient movement using the iGraph package in the R programming environment. Each node represents one location at VUMC and each directed edge represents a request for directions from one node to another. The size of nodes was proportional to the number of requests made from the given location while thickness of the edge represented the number of requests made from origin to destination.
To determine the types of locations that patients traveled between, we manually assigned each location to one of eight categories: Clinical (including procedure areas), inpatient unit, administrative (i.e. patient records, and information desks), public (i.e. cafeterias and gifts shops), parking, bathroom, elevator, and hallway. We then constructed another graph that illustrated travel between area types instead of specific areas. Next, we investigated which clinical areas were most commonly traveled between by examining which clinical nodes had the most edges between them. We also looked into instances where patients requested directions to areas where they were already standing according to BLE. Finally, we performed an analysis of the requests for directions from parking lots to explore whether patients parked in the parking lot closest to their destination.
Results
From September 1, 2016 when the WalkWays system was first implemented at VUMC to September 30, 2017, there were 3493 requests for directions from the application. Figure 3 shows the number of daily requests over the duration of the study. The number of wayfinding requests remains consistent throughout the study period with a large spike of 80 requests made on October 24, 2016.
Figure 3.
Number of wayfinding requests by day over the course of the study. Users made 80 requests on October 24, 2016.
A high level view of the types of areas patients travel between in Figure 4 shows that patients most frequently request directions from clinical areas to public areas and between clinical areas. Patients did not request directions to elevators or hallways nor do they request directions from bathrooms or inpatient units. Users request directions more frequently from parking areas and administrative areas than to these areas.
Figure 4.
Directed network graph of patient wayfinding requests by area type. Inner circle radius proportional to requests made from location. Outer circle area based on requests made to that location. Thickness of edges are proportional to number of requests.
A more granular visualization of the patient travel network in Figure 5 shows that patients most frequently requested directions from parking garages and to the courtyard cafeteria. Table 1 shows the most frequent origins and destinations requestsed. The most common clinical location to make a wayfinding request was the emergency room while the most common administrative area was guest services. Patients most frequently sought directions to the VUMC eateries and bathrooms.
Figure 5.
Undirected graph for all areas. The six most common origins are labeled. Radius of nodes are proportional to the number of requests made from that location. Thickness of edges is proportional to the number of requests. The arrangement of nodes is random.
Table 1.
Most frequent origins and destinations.
| Most frequent origins | Most frequent destinations |
|---|---|
| East garage | Courtyard cafeteria |
| Central garage | Au Bon Pain |
| Courtyard cafeteria | Women’s bathroom |
| Adult ER | Bistro on 8 |
| Au Bon Pain | Suzie’s café |
| Guest Services | Men’s bathroom |
Among 227 requests between clinical locations, only six routes had three or more requests. Patients of the orthopedic departments were the most active users of the wayfinding application. The most commonly traveled route from the plastic surgery clinic to the orthopedic rehab clinic was about 940 feet long. The average distance traveled for all wayfinding requests was 548 feet.
Patients and visitors made 406 requests from the two main parking garages to locations in the medical center. Among these, 240 requests were made from the East Garage and 166 were from the Central Garage. Assuming patients parked at the location where they requested directions from, approximately 62% of patients parked in the garage closest to the requested destination.
Finally, 174 requests were made to locations where BLE determined patients were already standing. This occurred most frequently at the Courtyard Cafeteria, the digestive disease center, and an internal medicine clinic.
Discussion
With wayfinding data, we were successful in identifying some trends in the movement of patients throughout the medical center. However, there are several limitations in our study that keep us from making stronger conclusions about patient experience. First, we assume in our analysis that patients are the primary users of the WalkWays application. Since there are no identifiers attached to wayfinding requests, it is possible that medical center staff are using WalkWays to request directions to parts of the hospital that they are not familiar with. The influence of these non-patient requests should be minimal since staff typically know their way around the medical center. Another limitation to our study is that BLE tracking coverage did not include the Monroe Carell Jr. Children’s Hospital at Vanderbilt and its South Garage. Adding the children’s hospital in the future will allow us to perform analysis on the difference in WalkWays users between adult and pediatric patients. Although the South Garage serves primarily the children’s hospital, some adult patients may park there since it is closer to some adult departments such as the eye clinic.
Our use of perpendicular distance to estimate distance traveled may slightly overestimate the length of routes that patients walked. While most VUMC hallways are laid out perpendicularly to the coordinate grid, diagonal hallways and open spaces would give patients a shorter route to their desired location. Additionally, since our distance calculations did consider changes in floors (VUMC consists of several multi-story buildings), patients may have had to backtrack to reach elevators before traveling to their destination. In future work, we will apply the WalkWays routing algorithm to calculate walking distance based on the recommended path provided by the application.
The biggest limitation to being able to infer associations between areas of the medical center is our small sample size. With 3493 requests over 13 months, there are fewer than 10 requests on average per day. The small number of requests means that a few patients that use the application frequently can have a large influence on the overall results. The high number of requests to and from the orthopedic clinic in Table 2 may be the result of good promotion of the WalkWays application from that unit, or a couple orthopedic patients that always use WalkWays. We are making considerations to include a non-descriptive identifier to each device that requests directions. This way, we will be able to see how many different patients are requesting directions on a given day. It will also allow us to determine whether patients make requests for multiple locations during their visit to VUMC. Putting together the “chain” of movement could provide richer insight into the overall patient experience.
Table 2.
Clinical areas most commonly traveled between and the distance between them.
| Origin | Destination | Distance in feet | # of requests |
|---|---|---|---|
| Plastic and cosmetic surgery clinic | Orthopedic rehab clinic | 940 | 24 |
| Orthopedic rehab clinic | Plastic and cosmetic surgery clinic | 940 | 9 |
| Radiology | Orthopedic rehab clinic | 542 | 4 |
| Medical center east surgery center | Medical center east pharmacy | 245 | 3 |
| Cardiac MRI | Radiology | 345 | 3 |
| Infusion clinic | Orthopedic rehab clinic | 1027 | 3 |
Despite these limitations, our analysis of wayfinding requests is benefitial to healthcare systems in several ways. Figure 5 shows that eateries such as the Courtyard Cafeteria, administrative areas such as guest services, and the emergency room are gateways to the rest of the medical center. Knowledge about where patients start their journey through the medical center could allow healthcare systems to strategically place staff or volunteers to help patients get to their destinations. Management could also use wayfinding requests as a surrogate for utilization. Historical or realtime wayfinding data could inform the allocation of resources such as cleaning for restrooms, relocation of wheelchairs, or maintenance for elevators that are more frequently used.
Hospital facility designers can use Table 4 to improve wayfinding signage in the medical center. Assuming patients are using WalkWays correctly and the application is functioning properly, the only reason why a patient would request directions to a location where they are already standing is that they are very near to their desired destination but do not realize it. Therefore, improved signage in these areas may help patients to know that they have arrived and thus reduce frustration. The results for optimal parking location in Table 3 could also enable medical center administration to improve patient experience. More than a third of patients are parking at the lot that is farther from their destination request. While some of these requests may just be a first stop before their clinical encounter (i.e. getting lunch at the cafeteria before your appointment), there is an opportunity for clinics to better inform their patients prior to their appointments on where to park to decrease walking distance. VUMC is in the process of including advance wayfinding directions in all patient appointment reminders that will specify which garage to park in and the indoor navigation route from the garage to the appointment. We expect this addition to the patient reminders to increase the sample size in our future studies.
Table 4.
Areas where patients made requests to when already standing at those locations.
| Location | # self-requests |
|---|---|
| Courtyard Cafeteria | 26 |
| Digestive disease center | 12 |
| Internal medicine suites 1 and 2 | 12 |
| Adult ER | 11 |
| Au Bon Pain | 11 |
| Occupational health clinic | 10 |
Table 3.
Number of requests made from VUMC East and Central garages by how far the garages were from the requested destination.
| Parked in East Garage | Parked in Central Garage | % Optimally Parked | |
|---|---|---|---|
| Destination closest to East Garage | 232 | 146 | 61% |
| Destination closest to Central Garage | 8 | 20 | 71% |
The information from Table 2 could also inform the placement of new or relocated clinics to minimize patient walking. If the trend of strong association between plastic surgery and orthopedic rehabilitation continues for the next few years, VUMC may consider moving these clinics closer together. The network of patient wayfinding requests provides data supported evidence for this operational decision by the healthcare system. More work needs to be done to investigate whether an intervention to decrease walking distance based on a recommendation from this analysis has an effect on patient satisfaction scores.
Data from this network analysis of wayfinding data can also help to inform enhancements in mobile applications that improve the patient experience. In the short term, application developers can improve WalkWays by suggesting commonly requested destinations based on their current location. This service could operate similarly to other recommender or autocomplete systems where options are presented to the user based on previously searched locations. Future medical center applications could also link indoor location data to the electronic medical record to guide patients to their appointment without having to request directions. Clinic staff could benefit from an integrated mobile application by viewing real-time locations of their patients. This application could allow staff to reach out to their patients if they are lost in the medical center, or make necessary modifications to the schedule if a patient is running late. Restaurants such as McDonalds, retailers such as Target16, and several airports17 are all using indoor positioning data to improve the customer experience and inform more efficient operations. Therefore, opportunities abound for the healthcare system to use mobile application data to understand and enhance the patient experience.
Conclusion
Using a novel data source, wayfinding requests from a mobile application, we were successful in inferring patient movement within the medical center and identifying some opportunities for improving the patient experience. The network of patient directions requests provides evidence to medical center management for the placement of clinics and the design of signage. Further development of mobile applications that enhance the patient experience may decrease patient wayfinding effort and increase efficiency of healthcare operations.
References
- 1.Ulrich R, Quan X, Zimring C, Joseph A, Choudhary R. 2004. The Role of the Physical Environment in the Hospital of the 21 st Century: A Once-in-a-Lifetime Opportunity. [cited 2018 Feb 6]; Available from: https://healingphotoart.org/wp-content/uploads/2012/11/UlrichPhyEnviron.pdf. [Google Scholar]
- 2.Ulrich RS, Zimring C, Zhu X, Dubose J, Hyun-bo S, Young-Seon C. A review of the research literature on evidence based healthcare design. Cent Heal Des. 2008;1(September):1–84. doi: 10.1177/193758670800100306. [DOI] [PubMed] [Google Scholar]
- 3.Devlin AS. 2014. Oct 31, Wayfinding in healthcare facilities: contributions from environmental psychology. Behav Sci (Basel, Switzerland) [Internet] [cited 2018 Feb 6] 4(4) 423-36: Available from: http://www.ncbi.nlm.nih.gov/pubmed/25431446. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Drainoni M-L, Lee-Hood E, Tobias C, Bachman SS, Andrew J, Maisels L. Cross-Disability Experiences of Barriers to Health-Care Access. J Disabil Policy Stud [Internet] 2006;17(2):101–15. Available from: http://journals.sagepub.com/doi/10.1177/10442073060170020101. [Google Scholar]
- 5.Bible JE, Shau DN, Kay HF, Cheng JS, Aaronson OS, Devin CJ. Are low patient satisfaction scores always due to the provider? Determinants of patient satisfaction scores during spine clinic visits. Spine (Phila Pa 1976) 2018;43(1):58–64. doi: 10.1097/BRS.0000000000001453. [DOI] [PubMed] [Google Scholar]
- 6.Zimring C, Zimring Craig. 1990. The costs of confusion: Non-monetary and monetary costs of the Emory University hospital wayfinding system. Georg Inst Technol. [Google Scholar]
- 7.Boulos M, Berry G. 2012. Real-time locating systems (RTLS) in healthcare: a condensed primer. Int J Health Geogr [Internet] [cited 2017 Apr 19] 11 (12) Available from: https://ij-healthgeographics.biomedcentral.com/articles/10.1186/1476-072X-11-25. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Barlow J, Wright C, Sheasby J, Turner A, Hainsworth J. 2002. Oct 1, Self-management approaches for people with chronic conditions: a review. Patient Educ Couns [Internet] [cited 2018 Feb 8];48(2):177-87. Available from: https://www-sciencedirect-com.proxy.library.vanderbilt.edu/science/article/pii/S0738399102000320?_rdoc=1&_fmt=high&_origin=ga teway&_docanchor=&md5=b8429449ccfc9c30159a5f9aeaa92ffb. [DOI] [PubMed] [Google Scholar]
- 9.Eton DT, Elraiyah TA, Yost KJ, Ridgeway JL, Johnson A, Egginton JS. 2013. A systematic review of patient-reported measures of burden of treatment in three chronic diseases. Patient Relat Outcome Meas [Internet] [cited 2017 Aug 2];4:7-20. Available from: http://www.ncbi.nlm.nih.gov/pubmed/23833553. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Graves BA. 2008. Jul 29, Integrative literature review: a review of literature related to geographical information systems, healthcare access, and health outcomes. Perspect Heal Inf Manag [Internet] [cited 2018 Feb 8];5:11. Available from: http://www.ncbi.nlm.nih.gov/pubmed/18698429. [PMC free article] [PubMed] [Google Scholar]
- 11.Cheng A, Levy M. 2017 2017. Determining Burden of Commuting for Treatment Using Online Mapping Services - A Study of Breast Cancer Patients. AMIA. Annu Symp proceedings AMIA Symp. [PMC free article] [PubMed] [Google Scholar]
- 12.Brunson JC, Laubenbacher RC. 2018. Feb 1, Applications of network analysis to routinely collected health care data: a systematic review. J Am Med Informatics Assoc [Internet] [cited 2018 Feb 7];25(2):210-21. Available from: https://academic.oup.com/jamia/article/25/2/210/4157164. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Steitz BD, Levy MA. 2016. A Social Network Analysis of Cancer Provider Collaboration. AMIA . Annu Symp proceedings AMIA Symp [Internet] [cited 2018 Feb 7];2016:1987-96. Available from: http://www.ncbi.nlm.nih.gov/pubmed/28269958. [PMC free article] [PubMed] [Google Scholar]
- 14.Ziio . VUMC WalkWays - Android Apps on Google Play [Internet] [cited 2018 Mar 8]. Available from: https://play.google.com/store/apps/details?id=io.ziiio.vumc&hl=en. [Google Scholar]
- 15.Google. Query Explorer - Google Analytics Demos & Tools [Internet] [cited 2018 Feb 27]. Available from: https://ga-dev-tools.appspot.com/query-explorer/ [Google Scholar]
- 16.Mittal S. 25 Retailers Nailing it with their Proximity Marketing Campaigns [Internet] [cited 2018 Feb 27]. Available from: https://blog.beaconstac.com/2016/02/25-retailers-nailing-it-with-their-proximity-marketing-campaigns/ [Google Scholar]
- 17.Babu P. 10 Airports Using Beacons to Take Passenger Experience to the Next Level [Internet]. [cited 2018 Feb 27] Available from: https://blog.beaconstac.com/2016/03/10-airports-using-beacons-to-take-passenger-experience-to-the-next-level/ [Google Scholar]