Hospital Access in New England

Lack of healthcare access is a popular talking point often discussed by many people, including but not limited to health care professionals, politicians, and the general public. Access has many different definitions, a major one being lack of insurance coverage despite being employed or a dependent of someone employed (Millman, 1994). Also frequently discussed is the impact of geographic barriers; distance to healthcare facilities can burden patients and lead to worse health outcomes (Buzza et. al, 2011). Because of the potential impact of poor health outcomes for individuals living far from a healthcare facility, I will be exploring the potential issue of lack of access due to geographic barriers on a regional scale. Additionally, I will take into consideration the quality of the healthcare facility and determine if there are any patterns of where poorly rated hospitals are located. I have decided to consider the hospital's quality because it may highlight other disparates regarding hospital locations other than distance. For example, are the highest quality hospitals only located in specific areas?

The hospitals' quality scores come from the US medicare website. The information is provided by the Centers for Medicare & Medicaid Services. Each facility is given an aggregated score out of five. This score is based on several quality measures such as readmission rate, patient experience, mortality rate, and timelines of care.

The demographic information, population, poverty levels, and race, come from the US Census. This information specifically comes from the American Community Survey. The demographic information is classified into five different brackets. The brackets were determined by natural breaks in the data.

The choice to make the hospital buffers 10 miles came from previous literature and studies. One example of such a study is How Far Americans Live from the Closest Hospital Differs by Community Type. There is evidence in the literature that the average person must travel roughly 10 miles.

These findings were reached by first selecting every census tract that had a hospital by using the select by location tool. I then exported this selection as its own layer. From there I ran a selected by attribute to determine which of these census tracts fell into the top two population density brackets. I then repeated these same steps with only highly rated hospitals.

These analyses were carried out by first selecting census tracts that intersect with the hospitals' 10 mile buffer. Then, after creating a new layer from the selection, a select by attribute highlighted which of these tracts are more densely populated. Once I completed this selection, I conducted another select by attribute but with the two least dense brackets.

These analyses were carried out by first selecting census tracts that intersect with the hospitals' 10 mile buffer. Then, after creating a new layer from the selection, a select by attribute highlighted which of these tracts are more densely populated. Once I completed this selection I conducted another, but with the two least dense brackets.

These findings were created by first selecting census tracts that intersect with the below average hospitals' 10 mile buffer. Then, after creating a new layer from the selection, I used select by attribute to determine which of these tracts are more densely populated. Once I completed this selection I conducted another select by attribute but with the two least dense brackets.

Many of the same results were found across all three levels of hospital quality. The majority of areas within 10 miles of a hospital are in the bottom two population density brackets. It is worth noting that hospitals with a below average quality rating were more likely to have a 10 mile radius intersect with a low population density census tract; 85% compared with 75% for average and 73% for above average.

I came to these conclusions after using select by location to identify all the census tracts with hospitals. Then, after creating a new layer from this selection, I used select by attribute to determine how many of these tracts had high and low rates of poverty.

These results were determined by selecting the census tracts that have areas within above average quality rated hospitals' 10 mile buffer. Then from those tracts, I selected the ones with high and low rates of poverty.

These analyses were carried out by first selecting census tracts that intersect with the average quality rated hospitals' 10 mile buffer. Then, after creating a new layer from the selection, a select by attribute highlighted which of these tracts are in the two lowest groups of poverty. Once I completed this selection I conducted another selection but with the two highest groups of poverty.

I came to these conclusions after using select by location to identify all the census tracts within a 10 mile span of below average quality rated hospitals. Then, after creating a new layer from this selection, I used select by attribute to determine how many of these tracts had high and low rates of poverty.

Based on these three maps, the results indicate that the majority of hospitals, whether they are highly or poorly rated, are located in areas of low poverty. Hospitals with the highest ratings, however, were more likely to be found in areas of low poverty and less likely to be found in areas of high poverty: 77% and 7% compared to 75% and 9% for average and 76% and 9% for below.

I came to these conclusions after using select by location to identify all the census tracts with hospitals. Then, after creating a new layer from this selection, I used select by attribute to determine how many of these tracts whose nonwhite population was greater than or equal to 43% and then less than or equal to 24%.

These results were determined by selecting the census tracts that have areas within above average quality rated hospitals' 10 mile buffer. Then from those tracts, I selected the ones with high and low percentages of nonwhite rsidents.

These analyses were carried out by first selecting census tracts that intersect with the average quality rated hospitals' 10 mile buffer. Then, after creating a new layer from the selection, a select by attribute highlighted which of these tracts are in the two lowest groups of population percent that is nonwhite. Once I completed this selection I conducted another selection, but with the two highest groups of population percent that is nonwhite.

These findings were found by first selecting census tracts that intersect with the below average hospitals' 10 mile buffer. Then, after creating a new layer from the selection, I used select by attribute to determine which of these tracts nonwhite population is greater than or equal to 43%. Once I completed this selection, I conducted another select by attribute for census tracts with nonwhite populations that are less than or equal to 24%.

All three of these maps have similar findings. The majority of hospitals and their 10 mile radii of any kind are found in majority white census tracts. Hospitals that are highly rated have the greatest discrepancy. 10% were in tracts with nonwhite populations greater than and equal to 43% and 77% were in census tracts with nonwhite populations less than and equal to 24%. These percentages compare to 12% and 76% for average rated hospitals and 13% and 76% for below average rated hospitals.

For each demographic category there were only small differences between the three quality levels of healthcare facilities. The greatest difference occurred when comparing hospital locations and populations. There was a difference of 10% and 8% between below average rated hospitals, average quality rated hospitals, and above average rated hospitals, respectively. The next largest discrepancy occurs in the racial demographic group of maps. There was a 3% difference between highly rated hospitals and poorly rated hospitals and their frequency of being located within a 10 mile radius of census tracts with 43% or greater nonwhite populations. Based on these analyses it would seem that there is a small and perhaps insignificant association between where hospitals of any quality are located and population density, poverty levels, and racial demographics. Based on all of my analyses, there does not seem to be a significant geographic barrier to access healthcare facilities. It would seem that public health officials should focus their efforts to improve access in different manners such as addressing the issue of lack of health insurance coverage, or rising prices.

There are, however, some limitations to this geographic analysis. One of the main limitations is that I did not account for the fact that there was not an even distribution of census tracts per attribute grouping. For example, there were more census tracts among the lowest two poverty rate groups than the highest two poverty rate groups. This means that looking at each map individually does not provide as clear a picture. If one were to only look at a map depicting where the highest rated hospitals are located within a 10 mile buffer compared to census tract poverty rates, one would be shocked to see so few located in census tracts with high poverty levels and may determine that there is a huge disparity. However, once you compare those findings to the map depicting where the lowest rated hospitals are located within a 10 mile buffer compared to census tract poverty rates, one can see that the percentages are very similar and therefore the disparity may actually be smaller than originally thought. Another limitation is that I was only able to select census tracts that had some areas that overlapped with the 10 mile buffers. This means that there are likely people I included as being within 10 miles to a hospital that are in fact further away. This is especially relevant in larger census tracts. Lastly, even though someone may live inside the 10 mile buffer, they may actually be further than 10 miles away. The buffer tool does not take into consideration the roads and paths so there may be people included as being within 10 miles of a healthcare facility that are actually more than 10 miles away.

Buzza, C., Ono, S. S., Turvey, C., Wittrock, S., Noble, M., Reddy, G., . . . Reisinger, H. S. (2011). Distance is Relative: Unpacking a Principal Barrier in Rural Healthcare. Journal of General Internal Medicine, 26(2), 648. doi:10.1007/s11606-011-1762-1

Millman, M. L. (1994). Access to health care in America. Washington, DC: National Acad. Press.

Minimum-Distance Requirements Could Harm High-Performing Critical-Access Hospitals And Rural Communities. (2015). Health Affairs, 34(4), 627-635. doi:10.1377/hlthaff.2014.0788