Neonatal Mortality and the Trade-off Hypothesis: A Conversation with Dr. Benjamin Sosnaud

Interview By Evan Hsiang

HHPR Senior Editor Evan Hsiang interviewed Dr. Benjamin Sosnaud, PhD. Dr. Sosnaud is an assistant professor of Sociology at Trinity University. His research centers around the inequalities in health outcomes, with his most recent work focusing on the socio-demographic inequalities of infant mortality. Dr. Sosnaud received his B.A degree at Duke University, and his M.A and Ph.D degrees at Harvard University.

To ensure clarity, the interview below has been minimally edited.

Evan Hsiang (EH): Thank you so much for taking time out of your day to speak with me. I wanted to first discuss your work on the Black-White gap in neonatal mortality using data from 33 states. Could you briefly explain the distinction between birth weight distribution and birth-weight specific mortality?

Benjamin Sosnaud (BS): Absolutely, it's a great question, one that I think is really helpful in understanding the nature of a lot of these disparities that I focus on in my research. Neonatal mortality is death in an infant’s first 28 days of life, so you can imagine that it is an extremely vulnerable period for a newborn. The fact that there are pronounced disparities between Black and white infants in terms of the likelihood of experiencing that mortality is certainly a real challenge for people like myself, who study health disparities and try to make sense of these patterns. It's definitely a really important example of racial inequality, so when you're trying to make sense of that type of disparity, the prior research in this area tells us it's helpful to account for two different processes that generate these types of gaps in mortality.

One of them is the fact that there's a difference in these babies’ likelihood of being born at high risk and in need of critical interventions. The way we've measured birth status is either, “Are you born at a very low birth weight?” or, “Are you born very prematurely?” The best data I have focuses on birth weight, but other people talk about premature birth in an equivalent way. You can imagine being born below the 2,500 gram cutoff for a normal birth weight, or even below 1,500 grams, which is the cutoff for very low birth weight, is extremely dangerous. One would need care from a neonatal intensive care unit with a lot of supervision. One of the factors that generates the disparities in neonatal mortality is the differences in the likelihood of being born in this high risk category, and it turns out that the factors that contribute to this likelihood of being born prematurely or the likelihood of being born very low birth weight actually have a lot more to do with the mother's experience while she is pregnant and even before her pregnancy. Exposure to factors like malnutrition, chronic stress, including the stress caused by racism and racial discrimination, can biologically send signals to the body that lead to premature, low-weight births. Therefore, the fact that there are racial disparities in the distribution of birth weights–the likelihood of being born in this low birthweight category–is one important contributing factor to racial disparities in neonatal mortality.

The other big factor is the fact that even among infants who are born in these high risk categories, there's differences in the mortality outcomes we see. We've already discussed the fact that there's a different likelihood of being born high risk between Black and white neonates, but there's also differential risk mortality if you are born high risk. Black infants have a higher risk of mortality if they’re born in the very-low birth weight category than white infants. If you have several infants in this extremely high risk category and some of them are having better birth outcomes than others, it’s really about the difference in the care they're receiving. Are they born in facilities that have the appropriate technology? Are they rushed there as efficiently and quickly as possible? Is the care that they receive sufficient for their needs? What's valuable about this distinction is that not only are we able to measure it but we're also able to relate it to different social processes. If I tell you that a big portion of the neonatal mortality gap is due to differences in birth weight distribution, you can ask, What’s going on throughout the lived experience of a mother during a pregnancy that causes her to deliver these very low birth-weight infants? If I tell you that there's a gap in disparity among infants of the same birth weight category, in this case very low birth weight, you can ask, What’s happening for them to receive differences in the quality of the care they need to survive being in this risky category? So I think that’s a really useful distinction to be able to draw theoretically and empirically.

There are some really great techniques to decompose the overall gap into these two different components. Prior research used this method with national level data, and they found that about 90% of the gap between Black and white neonates comes from these processes that result in differences in birth weight distribution. The other 10% are due to differences in birth weight-specific mortality. That insight led me to ask the question, Is the balance between these two contributing factors the same across all states? I had the hypothesis that it wouldn't be because states vary a lot, not just in the composition of the population, but also in their institutional structures, in the hospital systems, and in the way medicine is practiced and care is made accessible. So I wanted to do a separate decomposition in the 33 states in my sample where there's sufficient data to do the same analysis over this long period. We find that, like at the national level, the leading factor in all states is differences in birth weight distribution. Black infants are much more likely to be born in these extremely high risk categories, again, tracing back to lived experience and factors like racism and malnutrition. However, the degree to which differences in birth weight-specific mortality contributes to overall disparity does vary. In some states, birth weight-specific mortality doesn't matter at all, which tells us that racial disparities in neonatal mortality are driven entirely by the birth weight distribution. In other states, as much as 35% of the disparity is due to these differences in birth weight-specific mortality, which tells us that there's something going on in the medical system that produces different outcomes such that certain infants are being born at risk. Because we can isolate our focus on this very low birth weight category where the chance of survival depends so much on your access to intensive care and immediate post-birth interventions, we can ask another set of questions for future research: What is going on in these states where there are such profound gaps in access to this care? That's really what I hope is the value of my paper: setting up this really important focus. First, What is it about the lived experiences that is causing these differences in birth weight distribution? But also, What is it about the states where access to care seems to matter more, as measured by the differences in birth weight-specific mortality?

EH: As you mentioned, there are some states like Colorado where birth-weight specific mortality contributes significantly to this disparity. How do we implement policies to ensure that Black babies are supported by hospitals with the most up-to-date technology? How do we reduce the harms of implicit bias on the quality of healthcare provided?

BS: I think it's helpful to think about questions like that in different components. Part of the solution has to do with the overall possible medical infrastructure of the system: making sure that all the hospitals that someone might be born in have all the best technology and have specialists. Just as we have stratification and segregation in where people live, there’s stratification and segregation in which hospitals people use. Because my study is looking at all these states and making broad comparisons, I can’t exactly say what is the nature of the situation in Colorado or other states where these high percentages stand out, but certainly that's something that we could look at. The other fact that you mentioned is that even beyond the technology and the medical services, there’s the question of whether or not people are receiving the same care, and certainly the possibility of implicit bias is something we have to take seriously. Differential health outcomes occur when people are reporting symptoms but aren’t being listened to or when medical professionals aren’t as familiar with or have preconceptions about how care should be delivered for different groups. I think my paper reminds us of the importance of these factors. Greenwood et al. (2020) did some research in Florida showing that infants of color who were delivered by a physician of color actually had better outcomes than those who are delivered by a white doctor, providing evidence that there may be some implicit bias, which means that people aren't getting the same quality of care from the different providers that they interact with. Again, we're a bit beyond what the data in my paper speaks to, but these important questions tell us about the type of processes that we could potentially isolate, identify, and then come up with interventions to address.

EH: You also examined the trade-off hypothesis, which proposes that health inequity increases as overall under-five mortality is reduced. You ultimately found no evidence to support it in developing countries. Does this finding reduce the credibility of this hypothesis?

BS: I think it comes down to the context in which the research is investigated. The trade-off hypothesis has a lot of credible evidence especially in Western contexts. It’s the idea that as population health improves, sometimes gaps between social groups actually get bigger even though overall health is improving, and it's because, based on some studies, a lot of these technological advances in medicine are not equally distributed. When we come up with a great new treatment for diseases like cancer, those with the most access to health insurance, economic resources, and regular relations with doctors will find out about it first and then be able to afford it and apply it consistently. That's one reason why population health improvements, things that help health, are often associated with greater inequality. That trade off hypothesis does hold in Western contexts based on research in the US. In this paper that I co-authored with Harvard Professor Jason Beckfield, we wanted to know if we see the same trade-off in developing nations–nations that don't have the same robust public health infrastructure or level of institutional support over development in a lot of other contexts. We thought this was an important question to ask because the reductions in mortality that you're seeing in Ethiopia, Bangladesh, and other nations in the Global South are really dramatic improvements over a fairly short time due to institutional changes in public health dissemination and mass education. In contrast, mortality decline in the US refers to small improvements in health occurring over decades. So in this context of rapid change, Do we see the same type of trade-off that we saw in the US and other Western contexts? It's reasonable to have different hypotheses in response to that question. One hypothesis would be that we will see a trade off because just like in the US, we are making these investments such that those with more resources benefit more than those with fewer resources. The counter argument is that expanding mass education or public health infrastructure by building a new hospital or clinic may benefit those lower in the social hierarchy more than enfranchised groups that already had access to these resources. Maybe these interventions that are dramatically improving population health without causing increases in health inequity are doing so by producing more equitable distributions of resources than the type of interventions we see in the US. We collected data from a large sample of nations using the Demographic Health Surveys, used 20-plus years of data on child mortality as our outcome of interest, and collected data on educational expansion and public health investment. We found several really interesting patterns in the data. First, we do not see the same trade off in these nations, in the aggregate, as we do in the Western context. In this case, population health is improving and socioeconomic inequalities as measured by household wealth and maternal education are actually declining with child mortality over this period. The different contexts can help explain these different patterns. Second, we can draw an association between some of these specific drivers. We found that as access to education and public health investment expands in these nations, disparities in under-five mortality get smaller. So while we don’t have the level of data to directly observe if the same people who are experiencing these health improvements are also the ones getting access to education or public health infrastructure, we can at least say it would be consistent with the possibility that the nature of these institutional investments in public health and education seems to be the type of investment that actually is being accessed equally and is reducing disparities and improving health in populations at the same time. I think this is exciting because it shows us that the right type of intervention can benefit everybody while improving health for the population.

EH: Are there cases in the Global South where public health investments can exacerbate inequity as we frequently see in wealthier countries in the West? What are the qualities of negative interventions?

BS: Our model compared each country to itself and we pulled together the broad story from each country; it's unlike the previous study that I was talking about where we isolated each state. Thus, in a multivariate context I can't specifically answer that question, but I do remember that there were some nations where the story isn't as consistent, at least in a bivariate way. I can still speak to your question, which is, What is so special about these different types of interventions? I think here we can learn, again, a lot from the US context. Interventions that require reliable access to a physician might actually end up exacerbating disparities. Some people go to the doctor every year. It's easy to schedule an appointment, get regular check ups, or obtain new treatments. Others may not see physicians in time to treat their conditions since they are only able and willing to go to appointments in an emergency context. Thus, accessibility-dependent interventions can result in disparity. Interventions that require health insurance, where insurance doesn’t cover newer experimental treatments or doesn’t submit in every insurance plan in the same way, can block patients from life-saving treatment. Finally, inequities worsen when interventions do not account for the ease with which people can integrate them into their lives. If you have a job with long hours, you can't go to appointments at many times throughout the day. There's medication that doctors recommend not operating vehicles after taking, but if your job involves handling heavy equipment, that drug can't be easily integrated into the patient's life. In the same way, living in an unsafe neighborhood or working multiple jobs at night makes it harder to go workout or change one's fitness routine, so health interventions can be much more costly for someone who has less flexibility in their schedule. What's interesting about the nature of these broad interventions is that there is evidence that they can be implemented to benefit those who previously didn't have access to them before. Building a new clinic dramatically benefits those who had lacked transportation. Education is stratified in most societies, but building a new school, encouraging more women to go to school, and changing norms to support that are things that I think will help the people who are furthest removed from those interventions to begin with. That's where our theory came from. I really think our study sets up important future research on the ground work in these developing nations. What does it look like when you build a new school or clinic? Are the same people who were most removed from those access to the services before the ones who benefit? Our paper would suggest so, but I'd love to see some research that really fills in that gap and explores that more at the individual level.

EH: I would love to see that research as well. As you’ve mentioned, there are broader social inequities that contribute just as much if not more to differential health outcomes than the inequities created by healthcare systems themselves. How do we push for broader policy changes to address these social determinants of health when frequently, it seems like it's easier to rally public and government support for smaller alterations to the healthcare system?

BS: That’s a question that people in sociology, public health, and medicine are really starting to take seriously. People have begun to recognize that health is affected by how you live your day-to-day life, the neighborhood you live in–your environment beyond what happens when you actually show up at a doctor's appointment. One can receive medication, but oftentimes that's too late in the process, so I think that taking seriously the social determinants of health is a really good first step. Expanding what we mean when we talk about health policy is also vital to effecting change. I don't want to diminish the importance of expanding access to healthcare and insurance coverage, but that's just one thing. We can have health in all policies. Every policy we pass could be a health policy. Economic redistribution policies like a minimum wage increase are not just putting more money in some people's pockets. It also increases their ability to afford healthy foods. It's better for their health to live in a neighborhood where they're not exposed to pollutants. So that's an economic policy that is a health policy. Investing in infrastructure by making a new park is not just good for the community, but it’s also a safe place for people to be physically active. I think that's a really useful lens to remind us that we can have a health impact by pursuing a lot of different policies as opposed to some more narrow focus on just insurance coverage. Another way to communicate this health-in-all-policies framework whenever I talk to students, colleagues, or the public is the upstream versus downstream analogy. The idea is that if you see somebody drowning in a river and you jump in and save them, you made a big impact by saving their life. You could do that a lot, but at some point, you have to ask yourself why there are so many people drowning in the river and what is causing them to fall in the first place. Similarly, when the doctor sees patients with a certain illness, they obviously can save individual lives, but why are all these people showing up to the doctor? What is it about the neighborhood they're living in that is possibly making them sick again? We have to think back to exposure to chemicals or industrial waste, exposures that can accumulate in somebody's diet and lungs for a long period before they actually show up to the doctor with complications.Why is it that some people live in neighborhoods where there are no grocery stores that sell healthy organic fruits and vegetables whereas others can just walk to a place that has those services? Those types of upstream interventions that prevent people from showing up to the doctor in such high risk conditions in the first place are exciting. I think the framework resonates with people, and hopefully it sparks positive conversations about broader change.

EH: Where do you think the fields of health care, sociology, and health policy are headed in the next 10 years?

BS: Great work has been done to promote this framework, take seriously the study of health disparities, and make connections between health disparities that we see in these broader instructional frameworks. Jason Beckfield, as I mentioned, has been doing incredible work in that area. Jennifer Montez’s research is really groundbreaking in that regard. But I also think that as we gain a sharper understanding of theories, we now have the opportunity to start collecting some of the data that can tell this story in a more detailed way. Much of the data that I use in my study of health disparities has focused on mortality because we have collected data on who lives and dies through birth and death certificates. But there are other ways to assess someone's health status even throughout their life before they get to these points. I would like to see a fusion of those types of more nuanced individual level indicators. We can collect biomarker data that gives us a good sense of where somebody is in a broad set of health perspectives at different points in their life, maybe even with a longitudinal element, tracking them as they experience things. We also want to be able to locate people in a physical space so that we have not just a really clear indicator of how healthy they are but also what they are exposed to based on where they live. Many times in my research, I've narrowed the analysis down to the lowest level of geographic aggregation possible, whether it's the state or county, and still found myself dissatisfied with how local we could go. So I think all the great work done in the past decade has put us in a position in the coming decade to start collecting better data at the individual level, on the environment that somebody lives in, and hopefully, in an ideal world, on a longitudinal scale to track health with greater detail. I think we'll be able to really answer some of these questions that we can only kind of make guesses and hypotheses about based on the level of the data that we have so far.

EH: Thank you so much for giving us insight into the world of health care, sociology, and current health disparities.