Three Questions Can Help Quantify Health Disparities

This transcript has been edited for clarity.

Kathy D. Miller, MD: Hi. I'm Dr Kathy Miller, professor and associate director of clinical research at the Indiana University Simon Cancer Center. Joining me is Dr Samilia Obeng-Gyasi, assistant professor in the Division of Surgical Oncology at The Ohio State University in Columbus and one of my former colleagues.

Samilia, you have had an interest in health disparities and did a study on that using some of the clinical trial databases from our cooperative groups. Tell me about the study you just completed.

Samilia Obeng-Gyasi, MD, MPH: We looked at two ECOG-ACRIN Cancer Research Group clinical trials: E1199 and E5103. E1199 was a study looking at taxanes and adding it to originally established medications. E5103 was a study that added Avastin (bevacizumab) to established medications. We looked at these studies retrospectively to see if there was a difference in outcome among patients who had different insurance types at diagnosis, and also to see if there was a difference based on neighborhood socioeconomic status.^[1]

Miller: Insurance status is captured when patients enroll in trials at the National Cancer Institute. I think many people are not aware of that.

Obeng-Gyasi: The insurance status at diagnosis is captured; however, if it came just during the [clinical] trial, it may not be captured.

Miller: So, how was it broken down?

Obeng-Gyasi: We looked at the overall numbers of insurance within both studies. The majority of patients had private insurance and some patients had Medicare or Medicaid or were self-pay. About 85% of patients had private insurance and about 12%-13% had government insurance, meaning Medicare or Medicaid. We combined the Medicare and the Medicaid population to make the analysis a little bit easier because those numbers were smaller. But it was mostly private, government, and self-pay.

Miller: You also mentioned neighborhood socioeconomic status. How did you evaluate that?

Obeng-Gyasi: For neighborhood socioeconomic status, we used the Agency for Healthcare Research and Quality SES [socioeconomic status] index that used six variables: neighborhood poverty level, neighborhood wealth, occupation, income, neighborhood crowding, and education.

Miller: That is a way of essentially assuming that you are probably similar to your neighbors.

Obeng-Gyasi: Exactly. One of the strengths of doing it that way is it gives you a good idea about the kind of environment someone lives in. But one of the downsides is that not everyone who lives in a certain environment has the same resources. Neighborhood socioeconomic status is good because it gives stakeholders and institutions an idea of where people are coming from, but it may not always represent the individual's socioeconomic status.

Miller: In some towns, two blocks might make a very big difference in the environments and resources—for example, areas of gentrification. How small are the neighborhoods they are able to look at?

Obeng-Gyasi: We can go all the way down to census block, which is the smallest subgroup you can look at. Unfortunately, we could not get very granular because of the way the data were collected, so for this study, we looked at zip code and linked it to county-level information, which is a little bit higher than the census block.

Impact of Insurance Status

Miller: What impact did insurance status have on outcome?

Obeng-Gyasi: Insurance status did have quite a bit of impact on outcome. We found that patients with government insurance (Medicare or Medicaid) were less likely to complete a clinical trial and also had worse overall mortality than their privately insured counterparts. This is actually in tune with other studies among nonclinical trial populations.

Miller: It's similar to those data, but in some ways I think it was surprising to us. Because only 3% or so of adults enter clinical trials, we assumed, as a group, that they tend to have higher socioeconomic status, have more resources, and be more engaged in their health. I think we assumed that that might have been an equalizer.

Obeng-Gyasi: Right. Patients who have government insurance tend to be very specific populations because you have to meet certain eligibility criteria to get that insurance. Patients who have either Medicare or Medicaid may face certain barriers that privately insured patients may not have. For example, transportation may be a big issue. Among the elderly or people who are indigent, having access to a vehicle to go to multiple appointments may not be at their disposal. So, for those patients, even though insurance does not necessarily directly explain transportation, it might be a proxy for something like transportation.

Miller: Insurance provides coverage for the medical care, but there are still a lot of other barriers that it does not address, and those may be a bigger driving factor.

Obeng-Gyasi: Those are what we call social determinants of health. What insurance is doing here is acting as a proxy for something else. It's not the insurance itself that is causing the outcome but the other variables that contribute to qualifying for the insurance. Putting people in a situation where they require that type of insurance may be what is driving some of these outcomes.

Miller: The other challenging aspect is, these trials only give patients a certain portion of their treatment. Once their disease progressed or they finished treatment, maybe their access to care for ongoing follow-up care of other comorbidities is not the same.

Obeng-Gyasi: Right, that could have been different. Maybe the patient had access to a tertiary center where they had very high-quality care, but once they were no longer enrolled, the center that they might have transferred their care to might not have been as well equipped to help them in terms of their management. So, yes, there might be other variables that might have contributed to these outcomes.

Neighborhood-Level Data

Miller: What impact did the neighborhood-level data have?

Obeng-Gyasi: It did not have any impact. We discovered that neighborhood socioeconomic status did not show an association with trial completion and did not show association with overall survival. That being said, I should mention that the literature on area of residence—which is mainly what socioeconomic status signifies—and mortality has been very heterogeneous. Part of the problem is that when everybody does this kind of research, we all use different variables in our socioeconomic status index, so because there is heterogeneity in the indices, there is heterogeneity in the outcomes. The outcome is not so surprising and kind of consistent with what is out there. A possible explanation could be that although we're capturing data about the neighborhood, once again, it's not the individual's socioeconomic status and maybe that is why it's not reflecting a difference in outcomes.

Addressing Disparities

Miller: What do we do with this information to try to help our patients do better and to address some of those disparities that you have helped us see?

Obeng-Gyasi: We have to do a better job upfront of collecting social determinants of health in our data, and we need to collect it consistently throughout the treatment period. One example I give is how someone may start out with private insurance but then have a significant employment disruption, like losing a job, and switch to a different type of insurance (for example, Medicaid). Their outcomes may now change, but it will have the same insurance type at diagnosis. With that kind of patient, if you are collecting insurance status at multiple time points, you may recognize that they have gone from being somebody who was not in a vulnerable population to being in one. We also need to collect other variables like health literacy, transportation, childcare, and food insecurity—all things that may impact the patient's ability to continue participating in a clinical trial once they have enrolled.

Miller: A limited but consistent set of common data elements to address the financial and vulnerability issues could help us better understand how to intervene or identify patients who need intervention.

Obeng-Gyasi: That is exactly right. Once we are able to get more granular data points, and not too many of them, we can get a better understanding of how they actually interact with some of the clinical outcomes that we're seeing.

Miller: How do you screen your clinic patients for these sorts of vulnerabilities? These are really important topics. They also become a little sensitive, and I think we are a bit reluctant to ask patients about their financial information.

Obeng-Gyasi: In January at The Ohio State University in our surgical oncology clinics, we're going to start asking patients three questions that have to do with socioeconomic status. One will be about financial reserve, another one will be about food insecurity, and the other will be about transportation. We think that by collecting this data upfront, we'll be able to understand our patients' needs better and then be able to determine barriers that we may be able to help address at an institutional level.

Miller: I always love when we can boil things down to a short group of questions. And frequently, once you start unraveling them, you find a lot more information. Screening allows you to then deploy those resources in a more efficient way.

Obeng-Gyasi: Right, and because there are just three questions, it's very easy to have the nurses incorporate that in their history intake, which will make it easier to act upon the information we get.

Miller: Samilia, thank you for coming in and reminding us about the importance of these issues.

Kathy D. Miller, MD, is associate director of clinical research and co-director of the breast cancer program at the Melvin and Bren Simon Cancer Center at Indiana University. Her career has combined both laboratory and clinical research in breast cancer.

Samilia Obeng-Gyasi, MD, MPH, is assistant professor of surgery in the Division of Surgical Oncology at The Ohio State University in Columbus.

Follow Medscape on Facebook, Twitter, Instagram, and YouTube

Comments

Commenting is limited to medical professionals. To comment please Log-in.

Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.