So now I want to talk to you about this third article and get at this third question, does a racial composition of the neighborhood influence healthcare utilization? In here, what we're going to do is we're going to explore this third strategy, which is to look at this interaction between the disadvantaged status and where people are located. So to do this, we're going to look at another paper again by the same team, myself, Dr. Dinwiddie, Dr. Chan, and Miss McCleary. Here, this paper was published in the medical care research and review. It looks at disparities in healthcare utilization and the relationship between residential segregation. So we're interested in whether the disparities that we see from the use of physician services and non-physician services, whether that can be attributed to residential segregation as measured by the racial composition, in this case, it's the zip code of the residents. So what we do to do this analysis, is we take data from the 2006 Medical Expenditure Panel Survey, and we link it with data from the US Census that tells us the characteristics of the zip code in which the respondent is living in. Because we're using or accessing the geographic data from the Medical Expenditure Panel Survey, we have to conduct this data at the AHRQ research data center, AHRQ standing for the Agency for Healthcare Research and Quality. So we're looking at the utilization of roughly 17,500 adults who live in MSAs, for which we have a zip code level data for. Our healthcare measures of use are office-based physician visits, outpatient department-based physician visits, visits to a nurse, or nurse practitioners, or physician assistants, or nurse midwives, visits to other healthcare professionals such as chiropractors, optometrist, podiatrist, physical therapists, and then whether or not the person has unusual source of care. Our key independent variables of interests is the racial composition of the zip code that the individual lives in. So whether it's predominantly white, predominately African-American, predominantly Hispanic, or actually there's no predominant group, whether the zip code is an integrated zip code. So to do this analysis, first we do the traditional analysis that everyone does. So we ignore the geographic variables completely, and we just estimate the relationship between race and ethnicity in utilization. Just as an aside, we tried to do this for Asians, but we just had so few Asians in the medical expenditure panel survey data that we couldn't use the Asian data. So for those of you who are interested, that's why they don't appear in this analysis. So what you see here is the thing that people typically report. Blacks and Hispanics are using less physician services compared to whites, fewer office outpatient department visits compared to whites, fewer nurse and physician assistant midwife visits compared to whites, other healthcare professionals compared to whites, less likely to have a usual source of care compared to whites. You can see their significance level, in most cases they are significant, sometimes they are not significant, but their direction is lower utilization relative to white. So now what we did was, in order to get at our research question, is we've interacted the race of the respondent with the type of community zip code that they lived in. So whether they were a white person living in a white zip code, a white person living in a black zip code, a white person living in Hispanic zip code, white person living in an integrated zip code, we did that for blacks and Hispanics too. Then we computed relative odds ratios for these different groups. So what this table shows you is we use whites living in a white zip code as our reference group of interest. Then everyone else is compared to whites living in a white zip code. What we're interested in finding out is whether or not the type of zip code that you live in, does it disadvantage or advantage you relative to whites living in a white zip code. So what we see, the items that are circled in this chart tell us which ones are statistically significant. So we're presenting the data two different ways, one with a bar chart, and the other with the actual odds ratios are shown here. So what we see is that for whites living in white zip codes, that relative to whites living in black zip codes or Hispanics zip codes or integrated zip codes, whites living in white zip codes seem to have an advantage over whites living in Hispanic zip codes, and they have a lower odds in integrated zip codes, although it's not statistically significant. Surprisingly, whites living in black zip codes have a higher odds of using physician services. For blacks living in white zip codes, if we look at them relative to blacks living in black zip codes, blacks living in black zip codes have the highest odds ratio compared to whites living in white zip codes, as relative to blacks living in black zip codes or blacks living in Hispanic zip codes, or blacks living in integrated zip codes. So the way in which I would encourage you to look at it, you want to look at the table two ways. You want to look and see whether whites are advantaged relative to blacks regardless of where they live. So whites living in black zip codes, whites living in Hispanic zip codes, whites living in integrated zip codes. Then you want to see whether for a given race, ethnic group, whether living in the white zip code advantages you relative to your racial and ethnic counterpart living in another zip code. So we see this story. It's not completely the same, there are some differences. It looks like, for instance, that Hispanics are really just disadvantaged in living in integrated zip codes. It looks like blacks living in Hispanic zip codes are much more disadvantaged relative to blacks living in white zip codes or blacks living in black zip codes. But the world is complicated, it's not as clear as our hypotheses might suggest. But when you look at some of the other services, so this is for the non-physician services. So this is visits to a nurse practitioner or physician assistant, the relationships become much more pronounced. In this case, if you look at whites, regardless of where they live they are more advantaged than the blacks and Hispanics in those zip codes. But then as you look at blacks and Hispanics living in the white zip code, advantages them relative to blacks and Hispanics living in blacks and Hispanics zip codes. So this is more when we're thinking about the relationship between how the racial and ethnic composition influences health care utilization. It's not as strong for the physician care, but it is very strong for the other types of care that persons might seek from other providers. So in this case, it's the nurse practitioners and physician assistants, and this is for the other services. Again, we see the same thing, that whites are advantaged regardless of where they are, relative to blacks and Hispanics. However, persons living in white zip codes are advantaged relative to persons living in black, Hispanics and integrated zip codes relative to use of other health care professionals. So this is thinking about, again, what we're trying to do here is think about whether where you live, does it influence your access and utilization? So the way in which we would explain this is that we would say from our prior research that living in the black and Hispanic zip codes, you're less likely to have health care provider resources in those zip codes. So regardless of whether you're black, Hispanic or white, you have less access. That then also there seems to be a relationship within the zip code by race and ethnicity. Those things might have to do with some of the other factors that influence health care utilization, both in terms of person's ability to access and afford care, but other things such as the way in which the health care system interacts with whites relative to blacks and Hispanics. But this study is an illustration of how you can use the interaction between, in this case, our disadvantaged status and then also our neighborhood composition to explore the relationship between disadvantaged status and our outcome, which in this case is utilization. So what this study shows is that if we want to address disparities, we not only have to think about individual level factors that are associated with race and ethnicity and their influence on health care utilization, but we also have to look at community level factors that are associated with health care utilization and health care disparities, if we're really going to address disparities. So to conclude, we talked about three common strategies, one in which you're trying to stratify, and you're thinking about how in one context, race and ethnicity or the disadvantage status influences the outcome, and in another context how it might influence the outcome, and then you stratify. In the example I gave you, we were looking at an integrated environment versus the national environment. But you could think about it in the context of, we want to look at Medicaid recipients versus Medicare recipients, or persons who are in an HMO versus persons who are not in an HMO, or a person that are in a teaching hospital versus persons that are in a community hospital. You can think about it in that way. The second way is that we looked at how disadvantage status influence it as a main effect as it influence the outcome. So it is the independent variable of interests, and then we try to control for all the other factors that we think influence the outcome. So that we can then focus on the effect of the disadvantaged status on the outcome. Then the third way in which we looked at these variables, it's the interaction between the disadvantaged status and some other key independent variable. In the example we gave, the other key independent variable was the racial composition of the zip code, and then the disadvantage status with the individual's race and ethnicity. You could think of examples where that variable might be educational attainment. So the difference between race and ethnicity and persons of with less than a high school, high school, some college, and college educated, or you could think of that variable might be urban/rural. So you might be comparing, you have Hispanics who live in urban areas and Hispanics who live in rural areas, versus whites who live in urban areas and whites who live in rural areas. So there are many combinations that you can think of. But again, these are just three common strategies that people use. So these are readings that you can look at at your leisure, and here's my contact information if you're interested in more of my research and these methods and how to use these methods in applying to your own studies. You certainly should feel free to contact me or send me an email. Thank you very much for your time.
