[MUSIC] So now I want to look at our second question to illustrate our second strategy. So do blacks, and whites, and Hispanics have the same access to primary care? And in this case, this is a fairly straightforward strategy. We're looking at the relationship between the independent variable, which were going to look at, and our outcome, which is the availability of physician services. So this is a study that I did with Dr. Gniesha Dinwiddie, Dr. Kitty Chan, and Ms. Rachael McCleary. And it looks at the relationship between residential segregation and the availability of primary care physicians. It's a national study, so we're looking at the availability of primary care physicians in metropolitan statistical areas nationally. And the aim of this study is basically to see whether the availability of primary care physicians, whether it is the same in majority African American communities and majority Hispanic communities, relative to majority White communities. And do persons who live in such communities have similar access to primary care? So in thinking about this study, we really had two additional questions. So we wanted to know how does geographic access to primary care physicians for African Americans, Hispanics, and Asians compare to Whites? And then does the residential segregation reduce access to primary care for minority populations? And as I get further in this portion of the lecture, I'll try to help you make distinctions between what we mean by, first, a minority community, and then a minority community in the context of residential segregation for the broader MSA. So to do this analysis, we needed some data, first, from the AMA Masterfiles, and this is in the American Medical Association Masterfile. And what we did was we computed counts of physicians by specialty. And so for each zip code we know how many physicians in each particular specialty are practicing in that zip code. And then the other thing that we did was we used census data to look at for each zip code its racial and ethnic composition, as well as its socioeconomic factors, meaning its poverty level, education level, educational attainment levels, and so forth like that. And then the third thing that we got from the census is that at the MSA level, we also measured the degree in which the MSA was segregated. And the way in which you can think about this is one of the measures that we use is called the dissimilarity index. And the dissimilarity index tells you how many people from the minority group would have to move from where they're currently living to other places in the MSA in order to achieve a racially balanced MSA? So that every census tract in the MSA would have the same percentage of blacks and whites living in that census tract or the same percentage of Hispanics or same percentage of Asians. So that in that sense the community would not be segregated where blacks or Asians or Hispanics are all living in the same location. So the key variables in this analysis are measure of primary care physicians. And in this case, primary care physicians are those physicians who are family practitioners, general practitioners, internal medicine. And then because so many women rely on their OBGYN for some primary care, we included the OBGYNs also. And then what we did was we computed for each zip code, their population to primary care physician ratio. And if that ratio exceeded more than 3,500 persons per primary care physician, then we defined that area as a physician shortage area. And then we then categorized each zip code based upon the percentage of Black, Asian or Hispanics living in the zip code. And if a zip code had greater than 50% of its residents who are in the zip code were either Black, Asian, or Hispanic then we defined it as a majority Black or majority Asian or majority Hispanic zip code. Now racial composition by in Metropolitan Areas, I would say pretty lumpy. That's not really a technical term, but it sort of describes how the data looks. So most zip codes, not surprisingly, are majority White zip codes. However, 5% of zip codes are majority African American, almost 4% are majority Hispanic. Less than 1%, really like one-quarter of 1% are majority Asian. 80% of zip codes have fewer than 10% African Americans or Hispanic residents. And 95% have fewer than 10% Asian. So when we think about zip codes, for the most part, they're either really predominantly white to almost all white. Or we have these zip codes which are majority-minority zip codes, either majority black or majority Hispanic. Now, so why is this important? Because a good percentage of Blacks and Hispanics live in zip codes that are a majority of their race. So 40% of Blacks live in zip codes that are majority Black, and 37% or 38% of Hispanics live in zip codes that are majority Hispanic. Not so with Asians, Asians are much more likely to live in zip codes where they are not the majority of the residents. So what we're really interested in are these majority Black and majority Hispanic neighborhood zip codes, are they more likely to be physician shortage areas? So then we asked the question, what share of the population resides in a zip code that we would call a primary care physician shortage area? And the answer to that is 16.2%. That's too high, period. As a national number that's just too high. 13.2% of Whites live in such areas. Almost a quarter of Blacks and Hispanics live in such areas. And 9.6% of Asians live in such areas. So this is just a bivariate analysis, all we're doing is we're just looking at for Black residents, Hispanic residents, White residents. This is how we defined your zip code with regard to whether it's a primary care physician shortage area, and then what percentage of the population of the four groups live in such areas? We then asked the question, well, doing a zip code level analysis where the dependent variable in that analysis of whether or not a zip code is a primary care physician shortage area. Does the characteristics of the zip code influence the likelihood that that zip code is a physician PCP shortage area? So what here you have in front of you is the odds ratios from a logistic regression. Where in that regression what the dependent variable is a 0 1 variable, whether or not the zip code is a physician shortage area. And the independent variables of interest are, first, our race variables, which is majority Black, majority Hispanic, majority Asian. The reference group here is majority White. And then we also included other variables in the model, which included poverty status, percentage of the population that was near poor, meaning they're in that 100 to 200 % of the federal poverty line. And then we also included the educational attainment, the gender distribution, and the vacancy rate in terms of how many houses in the zip code are vacant. What you see in this analysis is that majority Black zip codes at 67% higher odds of being a physician shortage area compared to majority White zip codes. Hispanic, however, we find were actually 27% lower odds of being a physician shortage area compared to majority Whites. Now we're controlling for poverty, we're controlling for educational attainment, and we're controlling for gender distribution. And so we see that even though the 24% of the population lives in such areas, that it's not necessarily the fact that these communities are majority Hispanic. It may be other socioeconomic factors that are contributing to the likelihood that Hispanics are living in these areas without physicians. So this shows Blacks, Hispanics, the relationship for majority Asian is not significant, poverty rate is not significant. But the near poor rate, persons who are not poor but close to the poverty level, are more likely the higher the percentage of near poor people, the greater the odds of being a primary care physician shortage area. And then here is some more data looking at the educational attainment, and this is all in comparison to being a college graduate. We see that as you go from having some college to less than a college education, the greater those population the more likely the zip code is a primary care physician shortage area. So in this analysis what we see is that the odds of being a primary care physician shortage area increases for majority African American zip codes. And decreases for majority Asian zip codes relative to the degree of segregation here. So let me explain where these findings are coming from. So what we did was where we were interested in whether the relationship between the zip code being a majority African American, Hispanic, or Asian zip code, changed with the degree of segregation in the metropolitan statistical area. So if a zip code was a majority African American zip code, but in a city that was relatively integrated, relative to a zip code that was a majority African American zip code in a city that has a high degree of segregation. Does the likelihood of being a primary care physician shortage area, does it change for that majority African Americans zip code? What we find is that segregation increases the likelihood of being a shortage area, it also increases for African American zip codes. And then we find the reverse for Asian zip codes. What you see here in this chart is we have looked at five different ways in which to measure segregation. The dissimilarity index, the isolation index, the centralization index, the concentration index, and the spatial index. Earlier, I talked about the dissimilarity index and it measured what percentage of the population has to move in order to achieve integration. The isolation index tells you how isolated a minority population is relative to the white population. So the likelihood that on random, that a Black resident would encounter a White resident in the metropolitan area given where they live. The centralization measure, measures the degree in which the Black population is in the center city of the MSA. The concentration index and the spatial index, they measure in to some degree the amount of land area than which these populations are living on. Whether they're living on a relatively fewer acreage, if you will, of the property in the MSA relative to the Whites who live in the MSA. What you see in this chart, it doesn't matter the way in which we measure segregation. What you see in very low areas of segregation, and MSAs with very low degrees of segregation, the odds of being a physician shortage area for majority Black zip codes is close to 1. Meaning that in places where we don't see a great degree of segregation, we don't see this relationship between majority Black zip codes and a PCP shortage area. However, in communities, MSAs where there's a high degree of segregation, we do see a relationship. And so as the degree of segregation increases, the likelihood of a majority Black zip code also being a physician shortage area increase. For Hispanics, we don't pretty much see any relationship except for with regard to the isolation indices. And that shows us that as the degree of segregation increases, actually the likelihood of being a physician shortage area for Hispanic zip codes goes down. And then for majority Asian zip codes, we see again this negative relationship between the amount of segregation and the likelihood of that zip code being a physician shortage area. And in fact, when we get to high levels of segregation, the majority of Asian zip codes seem to be advantage relative to majority White zip codes, in that they're less likely to be physician or shortage areas relative to majority White zip codes. So we have some explanations why we see this in the data. One is that in minority populations, we're not controlling for insurance status in our measures. So, in particular, Blacks, African Americans, are more likely to be covered by Medicaid and Medicare. And so to the degree that they don't pay as well as private insurance pays, then physicians will be less likely to locate in communities where the residents are dependent upon Medicare and Medicaid. And this data predates the Affordable Care Act, and so there also is a high degree of number of uninsured residents too. So compared to communities where everyone has private insurance, relative to communities where a large proportion of the residents are covered by Medicaid or uninsured, physicians are not necessarily going to to locate in those areas. The other is is that we know that the percentage of minority physicians is relatively low relative to the percentage of the population. So in that sense, there are just fewer physicians of the same race or same ethnicity to locate in African American communities. So then what explains the Asian result? Well we know that we get a high number of our physicians who are foreign medical graduates. And these international medical graduates are locating in states with large Asian populations, California, New York, Texas, Jersey, Illinois, and Florida. And so say you're locating in a community and you come from the Pacific Rim. So you're coming from Japan or Korea or from China, and you locate in a community in California and you speak those languages. Being in a community that is Koreatown or Chinatown, in say Los Angeles, might actually be an advantage for you, because you might readily be able to speak the language and connect with the patient population there. And so when we see these location decisions by our foreign medical graduates or international medical graduates locating in these areas. It may in fact be an advantage for them, and that's why we might not see this relationship. [MUSIC]
