Comparing Health Outcomes in San Francisco and Boston Metro Areas

Abstract

In many tangible and intangible ways, the Boston and San Francisco metro areas are very alike. This paper investigates whether this resemblance between the two metro areas leads to similar health outcomes. Health outcomes, estimated at the level of census tracts and collected as a part of the 500 cities project are compared to assess this similarity. The named cities of Boston and San Francisco along with four other suburban cities from each metro area, which have nearly equal numbers of census tracts and populations are considered in the comparison. Based on the 13 health outcomes, census tracts from these ten cities are grouped into three clusters using k-means clustering. These clusters are labeled Poor, Fair and Good based on their mean prevalence of health outcomes. A quantitative assessment of the distribution of the census tracts among these three clusters shows that the Boston metro has a significantly larger percentage of tracts in the Poor cluster compared to the San Francisco metro. This is primarily because the percentage of tracts with poor health outcomes is higher in the city of Boston compared to the city of San Francisco. However, in the other suburban cities, where the populations have similar median ages, the distribution of the census tracts is very close. Geographical visualization, conducted by mapping the census tracts to the shape files from the U.S. Census Bureau, shows that in the cities in the Boston metro area, census tracts in a single cluster are more likely to be adjacent, whereas, they are more likely to be dispersed or scattered in the cities in the San Francisco metro area.