Numerous reports have predicted a huge population increase in Lane County, Oregon, especially in the Eugene-Springfield area. This creates a high demand for hospitals. There are five hospitals in Lane County currently with one in Florence, one in Cottage Grove, one in Springfield, and two in Eugene. According to the Portland State Research Center, Lane County has a 2015 population of 361,540. Lane County is projected to have a population increase of 16,258 totaling 377,798 in 2020. From 2020 to 2025, there will be an increase of 18,092 people resulting in an estimated total population of 395,890. Lane County’s population will reach 413,693 in 2030 after an increase of 17,803 since 2025. In this case, the total population gain from 2015 to 2030 will be 52,153. Of that 52,153-population gain, Eugene expects a 31,603-population gain, Springfield expects 10,277, Veneta expects 2,321, Cottage Grove expects 2,201, Florence expects 1,733, Junction City expects 1,630, Creswell expects 1,515, and smaller increases would occur elsewhere in Lane County. Will additional hospitals be needed before 2030? If so, how many and where should they be located? GIS analysis provides a suitable approach for addressing this problem because GIS has the potential to play a critical role in determining when and where to help health providers balance the location, type, and patient workload of physicians in a network, as well as having incredible spatial visualization power (ESRI). Location-allocation is a great way to solve an issue such as this because it “involves two types of decisions: where to locate and how to allocate demand for a service” (Longley et al., pg. 322).
Methods
First, I gathered datasets from the Oregon GIS Clearinghouse that would be used as my variables for the project. Those datasets included Oregon counties, Oregon census tracts, Oregon zones, hospital locations, and networked street data. Next, I developed a model on a piece of paper that would be used as a guide for future steps and analysis. Lane County was my study site so I had to extract Lane from the Oregon counties dataset. The same process of extracting Lane County was used on the Oregon zones dataset, Oregon census tracts dataset, and hospital locations dataset. Now that I had a map layer for each variable within my study site, Lane County, it was time to give each layer the same map projection and resolution. This step was important because inconsistencies and errors may arise if one or two maps had a different projection or resolution than the others. I used a resolution of 2,087 feet by 2,087 feet, or 100 acres. I chose Lambert Conformal Conic as the map projection because it keeps area consistent and there is no distortion. Next, I reclassified my Lane County zones dataset because I wanted four land use types instead of the 31 provided. Any land use types with descriptions of commercial, office, campus, industrial, public use, high-density residential, or medium-density residential were assigned as urban in the reclassification. Land use types with descriptions of farms, minerals and aggregate were assigned as agriculture in the reclassification. Land use types with descriptions of forest, park, estuary, or open space conservation were assigned as forest in the reclassification. Land use types with descriptions of rural residential, low-density residential, rural commercial, or rural industrial were assigned as rural residential in the reclassification. Then I exported a map showing these four land use types. While keeping that map saved, I performed another reclassification but this time I just wanted two land use types, urban and non-urban. Non-urban land use combined the agriculture, forest, and rural residential land types from the previous step. Using just the urban land type creates an easy transition into the next step because the next focus was urban land use expansion. Based on the research I gathered on Lane County population growth, I calculated that only .47% of the overall population increase from 2015-2030 would be outside of urban zones. It was clear that urbanization was at play in this projection. Assuming outward expansion, I created three time steps that had a 2,087 feet increase per time step away from the original urban zone of 2015. I used a constant expansion each time because the population growth rate was fairly consistent from 2015-2020, 2020-2025, and 2025-2030 so the urban growth boundaries would have expanded at very similar distances if I had adjusted the distances accordingly by their actual five year growth rates anyways. This process draws the new urban zone for 2020 on the map, which adds a distance of 2,087 feet outward from all urban spaces on the 2015 map. I repeated this step for 2025, except the distance added another 2,087 feet because that was the five-year distance expansion for each step, therefore the edge of the 2025 urban zone was 4,174 feet. Once again, another 2,087 feet was added for creating the 2030 urban zone, which had an edge of 6,261 feet away from the edge of the original 2015 urban zone. Now I four urban zone maps to use for the final step. I had to determine which hospitals were best able to serve the population of Lane County over given travel times to hospitals. To do this, I used a location-allocation model based on the networked street map, hospital locations, and urban zones representing demand points. I performed this step for each scenario; the urban zone in 2015, 2020, 2025, and 2030. I knew Lane County expects a population increase of 52,153; therefore, I divided that estimate by three since there were three urban growth time steps used in the analysis. Every step, outside of the data collection, was accomplished in ArcGIS.
Results
Figure 1 displays the Lane County Land Use after the first reclassification process. Figure 2 displays the Lane County Land Use after the second reclassification process that isolated urban zones from non-urban zones in 2015. The urban zone expanded outward in all directions by 2,087 ft away from the 2015 urban zone as seen in Figure 3. Urban cells continued to get larger after simulating urban expansion in 2025 as seen in Figure 3. The urban zone expanded one more time after the 2030 simulation as seen in Figure 3. With the assumption that 99.6% of population growth starting in 2015 would occur in the urban clusters on the map, creating new urban clusters in the forested areas was not necessary. After running a location allocation model for each time step, which represented intervals of 5 years after the default year of 2015, the demand points, which represented urban population zones, increased significantly after each time step concluded. There were 571 demand points for 2015 after the first time step concluded. After the second round, which simulated 2020, there were 924 demand points. After the third round, which simulated 2025, there were 1295 demand points. For the final round, which simulated 2030, there were 1784 demand points. The lines shown in each map represent the recommended hospital for each demand point based on travel time. Figure 4 displays the results from the location allocation model for the 2015 urban zone demand per hospital. Figure 5 displays the results from the location allocation model for the 2020 urban zone demand per hospital. Figure 6 displays the results from the location allocation model for the 2025 urban zone demand per hospital. Figure 7 displays the results from the location allocation model for the 2030 urban zone demand per hospital. In other words, figures 4 through 7 show the spatial allocation of hospital services based on growing demand over time. Figure 8 is a bar graph showing the amount of demand for each hospital in Lane County from the 2015 location allocation results. Figure 9 is a bar graph showing the amount of demand for each hospital in Lane County from the 2020 location allocation results. Figure 10 is a bar graph showing the amount of demand for each hospital in Lane County from the 2025 location allocation results. Figure 11 is a bar graph showing the amount of demand for each hospital in Lane County from the 2030 location allocation results. As you can see, the Sacred Heart Medical Center at the University District serves at least twice as many demand points as the other four hospitals. The Sacred Heart Medical Center at River Bend and the McKenzie-Willamette Medical Center will both gradually become overcrowded as well by 2030. I recommend two new hospital locations by 2030 that will help balance the travel times to hospitals, the growing demand because of population increases, and the distance away from current hospitals. The first new hospital should be built in the southern part of Junction City because residents are already traveling 15 to 20 minutes to get to the nearest hospital in Springfield and Eugene due to not having their own local hospital, and Junction City is projected to be one of the fastest growing cities between 2015-2030. As a result, residents in Coburg, Veneta, and the northwest areas of Eugene would be presented with another hospital option as well, depending on if the proposed Junction City hospital is closer than the current Eugene & Springfield hospitals. The second new hospital proposal would be located in southern Creswell. This location would give all Creswell residents a local hospital and decrease travel times for residents in Oakridge, Westfir, and the southeast corner of Eugene as well. This would help decrease the overwhelming demand at the Sacred Heart Medical Center at the University District. Creswell also expects a large population growth, which increases the demand for a more localized hospital.
Figure 1:
Figure 2:
Figure 3:
Figure 4:
Figure 6:
Figure 7:
Figure 8:
Figure 9:
Figure 10:
Figure 11:
Conclusion
The results demonstrate an imbalance between facilities and their respective demand. Numerous studies have already been done especially by the Portland State Population Research Center. They have developed a population forecast for Lane County in two separate studies, one for 2015-2065 and another from 2008-2035. They base their results on growth rates, fertility rates, mortality rates, and net migration. The City of Eugene also develops urban growth boundaries, but rather at a more conservative level than in the latter studies. There were a few possibilities for uncertainty and error. The first possibility is the conversion from vector to raster datasets. It is possible that for example a forested cell may not be in the same spot. Another uncertainty could have been when I decided to include rural residential as a non-urban zone. In 2030, the urban expansion since 2015 will be 6,261 feet from the edge of the current urban zone and 99.6% of the 52,153-population increase will reside in this space. This places about 43 people per cell of 100 acre space, which is a fairly realistic, yet conservative result. Finally, Lane County will need to provide two additional hospitals. One will be located in Creswell while the other will be located in Junction City.
References
Portland State Population Research Center. Coordinated Population Forecast 2015-2065. Lane County Urban Growth Boundaries (UGB) & Area Outside UGBs. Retrieved from https://www.pdx.edu/prc/sites/www.pdx.edu.prc/files/Lane_Forecast_Report_201506.pdf
Population Research Center College of Urban and Public Affairs Portland State University (2009). Population Forecasts for Lane County, its Cities and Unincorporated Area 2008- 2035. Retrieved from http://www.lanecounty.org/Departments/PW/LMD/LandUse/Documents/Forecasts_Report_Final.pdf.
ESRI. GIS for Health Care Today and Tomorrow. Arc User. Retrieved from
http://www.esri.com/news/arcuser/0499/umbrella.html.
Longley, P.A., Goodchild, M.F., Maguire, D.J., Rhind, D.W. (2015). Geographic Information Science and Systems. Fourth Edition. John Wiley & Sons Inc.
City of Eugene. Eugene Zoning Map. Retrieved from