Rachel Mayer on the Fight Against Maternal Mortality

Cameron Boozarjomehri (Left) Rachel Mayer (Right). Photo: Adam Barnes

Interviewer: Cameron Boozarjomehri

Welcome to the latest installment of the Knowledge-Driven Podcast. In this series, Software Systems Engineer Cameron Boozarjomehri interviews technical leaders at MITRE who have made knowledge sharing and collaboration an integral part of their practice.

Rachel Mayer has grown up with medicine on the mind, but one subject has always been at the forefront: Maternal mortality. In a country as advanced and capable as the United States, why is Maternal Mortality still so high? While many public health practitioners often turn to medicinal interventions, she turned to data. Listen in as Rachel guides us through how her own work in the MITRE Innovation Program is helping save mothers and children everywhere.

Click below to listen to podcast:

Podcast Transcript
Cameron:	00:15	Hello and welcome to MITRE’s Knowledge-Driven podcast, a show where I, your host, Cameron Boozarjomehri, have the incredible fortune of interviewing brilliant minds across MITRE. Today I’m joined by the ever-talented Rachel Mayer, who is going to tell us about a fairly morbid but incredibly important subject plaguing America right now and how MITRE is working to help overcome this problem. Rachel, would you like to introduce yourself?
Rachel:	00:37	Thanks, Cameron. As you said, my name is Rachel Mayer. I’m a senior health systems analyst at MITRE, and I’m the principal investigator on our maternal mortality research project. Why I came to MITRE and why I’m very fascinated in this topic is that I’ve always been very driven towards public health and solving some of our issues that are facing the U.S. healthcare system as a whole. I grew up surrounded by clinicians, and I was always intrigued to ask those tough questions and find solutions.
Rachel:	01:13	I feel very fortunate that MITRE is investing the time and resources to tackle this important issue and that I get to dedicate my work time through the MITRE Innovation Program to research and try to help solve this tough challenge.
Cameron:	01:30	Thank you so much for joining us. Yeah, this is actually something that I’ve been hearing a lot about actually in different parts of MITRE. I’m sure anyone who’s listened to previous episodes will know the MITRE Innovation Program is just an ongoing project where anyone who has a good idea that we really think will help the world can come and pitch the idea and get it funded. I’m glad to see, especially for this problem, that you were able to get funded. So with that in mind, why don’t you give us a little more background on exactly what maternal mortality is and what this bigger problem you’re trying to address is?
Rachel:	02:02	Sure. According to the CDC’s definition, a death can be classified as a pregnancy-related death while the woman is pregnant or up to one year postpartum. There is a varying definition by the World Health Organization that defines pregnancy-related death as the death of a woman while pregnant or up to 42 days postpartum. So it’s important to note those differences, and we’re using CDC’s definition, as well as do most entities within the U.S.
Rachel:	02:33	The United States is the only developed country in the world with an increasing maternal mortality rate, and within those rates exist prominent racial disparities. Non-Hispanic Black women are three to four times more likely to suffer from a pregnancy-related death than white women. Those racial disparities have been consistent for a few decades now, despite interventions reduce them. What we started out with is really trying to understand the current state of maternal care programming in the U.S. Why has the U.S. maternal mortality rate been increasing, and what can be done to reduce it?
Rachel:	03:14	California is having the most success in this space. They stood up a Maternal Mortality Review Committee, where a team of interdisciplinary experts came together to determine the causality behind these pregnancy-related deaths and if they were in fact preventable. Nationwide, around 60% of these maternal deaths are in fact preventable, which means that some intervention could have been taken to save that woman’s life. These committees determine the causality and the preventability, and then California specifically designed these toolkits to address the leading causes of pregnancy-related death. They implemented those toolkits across the state of California, and within seven years of implementing these toolkits, they reduced their maternal mortality rate by 55%, which is really amazing when you’re considering the fact that everywhere else in the country, it’s increasing.
Cameron:	04:10	With that in mind, I’d actually like to unpack a lot of what you just said. I think the first thing I’d really want to understand is …the American medical system is full of so many different actors and factors, I imagine it was really difficult for them or anyone to really find the data that would even hint at what any solution might be. Do you think you could speak a little to how they came up with the data and how they were able to work through that process?
Rachel:	04:34	It’s important to note the difference between pregnancy-associated death and pregnancy-related death. If a woman is in a car accident and… while pregnant and dies, but dies because she was in the car accident but not necessarily because she was pregnant, regardless of the situation, on her death certificate, it would have been marked that she was pregnant when she died, and she would be pulled into a sample of women that would then be reviewed to see the causality of her death. But an example of a pregnancy-related death is if a woman went into labor and she suffered from obstetric hemorrhage and unfortunately died. She died because of a specific condition because of her pregnancy. With the addition of this pregnancy checkbox on a death certificate, which is how the CDC or any organization pulls to figure out which women actually suffered a pregnancy-related death, all of those women are grouped together, and so it takes a lot of effort to sort through all of those women to figure out which ones are actually pregnancy-related instead of pregnancy-associated and determine the causality of her death.
Rachel:	05:37	Once that Maternal Mortality Review Committee determines that, it’s reported to their state level. It’s actually only voluntary to then report that to the national level. Different states are using different definitions, have different experts within their teams. There’s not a lot of consistency between how states are classifying it, there’s a lot of confusions. These pregnancy-related checkboxes, as well as the ICD-9 to 10 codes in 2015, has caused a lot of concern in terms of what actually are the rates. Especially in Texas, they actually accidentally inflated their rates and so they had to go back to re-identify if their numbers were in fact accurate.
Rachel:	06:29	So there’s a lot of inconsistencies, and in the last two years we’ve been exploring those data quality issues, but there’s a huge need for overall just more of an emphasis on how it’s defined, how it’s collected, how it’s stored, and how it’s disseminated for people who are trying to initiate change in this space understand what’s really happening. Because if you’re designing programming and you’re not looking at the data, how do you know you’re solving the right problem?
Cameron:	07:00	And even then, I mentioned it was hard to normalize that data because they could be reporting all sorts of reasons why things might be happening. Being able to distinguish between what in one state might clearly be a pregnancy-related death, another state is just a pregnancy-associated death, getting to be able to parse those two out will be very complicated across all 50 states.
Rachel:	07:21	Yeah. When you’re deeming off that pregnancy checkbox, there’s a lot of work that has to go into that. It’s more than looking at the coroner’s report. Some states are even going to social media, they’re going to anecdotes, they’re talking to family members to really determine what the causality of that death was and if in fact her death was preventable, because there’s a lot of different factors that can interplay into maternal mortality rather than, you know, X woman hemorrhaged. It’s a very complicated topic that deems across so many different public health issues.
Cameron:	07:56	And so from there, how did they take that data and in California come up with these kits?
Rachel:	08:01	In their Maternal Mortality Review Committee, they collected the data for X period of time and then sorted it to find out what their leading causes of pregnancy-related death were, and in terms of if the deaths were preventable, what interventions could have been done to prevent them, and they incorporated all of that into these quality improvement toolkits that addressed those. So they understood their population, they knew what was needed to correct those wrongs, and they implemented them and had success. I think the country as a whole needs to learn from that and apply that methodology to all the other states.
Cameron:	08:43	Maybe before we move on to MITRE’s role, you could give us a little more about what we might expect in this toolkit, because it’s not a literal toolkit.
Rachel:	08:50	Right. Each one of them are basically a PDF that can be given to a hospital or a clinician, and it includes educational training, clinical protocols. Basically, if it’s addressing pre-eclampsia, what are all the things that should be known about pre-eclampsia, what education should be surrounding that, and what interventions should be put in place to effectively treat it.
Cameron:	09:13	With all that in mind, where does MITRE fit into all of this?
Rachel:	09:17	Sure. A team of different federal agencies and organizations within the obstetrics and gynecology and maternal health space have teamed up to implement a program called the AIM bundles, which has toolkits of similar nature to California’s programming, that they’re trying to get different states to apply within their hospitals, but it’s done on a very different scale and it’s optional to institute these bundles. It’s too early to see the effectiveness of them. It’s done on such a finite scale of one hospital is doing it in this city, and one hospital is doing in that city, so we don’t understand yet the effectiveness of it. It’s very hard to understand the data behind that and apply it to other states, versus California has more numbers and data on the effectiveness of the toolkits and what would happen if it was widely dispersed and implemented.
Rachel:	10:21	Our MITRE research team developed the MITRE Maternal Mortality Interactive Dashboard, which we’ve coined us 3MID, 3 M-I-D. What it does is, in its nature, it’s an interactive dashboard. It allows the user to develop these what-if scenarios to see how different factors interact with one another to influence maternal mortality rates. .
Cameron:	10:47	In a dashboard like that, and it sounds like you would have a lot of representative data, how are you making sure that you don’t accidentally let anyone’s individual information get leaked? How do you protect the privacy of these women who are suffering?
Rachel:	11:01	Sure. We’re using a synthetic population, which is a statistically significant reproduced population of a dataset.
Cameron:	11:09	Just to be clear, that means that it’s not the same dataset, but it is close enough that anything gleaned from it is the equivalent of what would have been gleaned from the actual populace.
Rachel:	11:19	Right. We have a sample size of around three million women. Within each simulation, we pull a thousand of those women that represent the dataset in which we’re pulling the data from. We have the demographic factors of those women, so their race, age, income. It calculates their risk score of their potential to suffer a maternal death, and taking into account their county in which they live in.
Cameron:	11:53	I realize the obvious goal here is we want to cut down on maternal mortality, but as you roll out this dashboard and give other people access to it, what are you hoping users would glean from these tools?
Rachel:	12:04	Sure. By creating these what-if scenarios, we hope that policy makers can see how to most effectively allocate their resources. The successful implementation and use of these toolkits requires a lot of resources, and resources includes money. To know what areas are in the most need and to allocate those resources to the areas of most need to have the most effect, to impact the most lives, that’s the goal, is to know how to be smart and promote data-driven decision making.
Cameron:	12:39	Yeah, so it sounds like there’s really three things you’re achieving here. The first and most difficult is just showing that there is this problem, and that is a problem of getting good data, having it categorized, appropriately representing a problem or at least identifying what the pattern is. Then the second is how do you turn that data into a meaningful, actionable way of doing anything, which it sounds like the toolkit solves. This last point is we need the dashboard to say, “All right, we did the footwork. We found this is the data in California, this is a toolkit from California.” Now, is that data limited to California? If I’m in Florida, if I’m in Ohio, if I’m in Alaska, can I still glean similar ideas and results in terms of understanding who is underserved in my constituency or in my town, and being able to turn that into something actionable that helps those people out?
Rachel:	13:33	Yeah, that’s a great question. Right now, in 3MID, we just have California and Georgia. We have the data from those two states in relation to what’s necessary to run these simulations, but the end goal is to have every single state. But as you can imagine, every state is different, they face different issues. It’s important to have the specific data to take into account infrastructure and culture and rural versus urban. There’s so many complicated factors that are contributing to a specific state’s public health concerns that ultimately contribute to maternal mortality rates.
Rachel:	14:18	We’re working right now to set up partnerships to obtain more data so that we can have a tool with more states within it, but it takes time. Ultimately, I mean, that’s the goal, is for every state to be able to use this dashboard to understand how to reduce their maternal mortality rates.
Cameron:	14:39	Can you walk us through an example of how this tool has been, or might be, used?
Rachel:	14:43	Sure. If a decision maker of how funding is going to be allocated is trying to determine, “Hey, I have a million dollars. I want to help maternal health programming. I want to help reduce maternal mortality rates in my state, because our rate is really high. How do we do that?”
Cameron:	15:00	Yeah. This could be a charity, it could be pretty much anyone who is in any position of power to make a decision like this.
Rachel:	15:06	Right. They need to understand their environment, they need to understand where the most need is and where their dollar spend would have the most impact. So they could use this tool to figure out within their state what counties are in the most need and the population within that county, because different maternal health conditions affect different races in different capacities.
Rachel:	15:32	For example, pre-eclampsia affects Black women at a higher rate than white women. Say you find a county within your state that has a high black population that suffers from pre-eclampsia at a very high rate, you could say, okay, we understand that they should be using this toolkit. If we applied this toolkit to this population, given this dollar amount that we have, what sort of effect could this potentially have, and how does that compare to the effect in a different county? It’s not about which county deserves it or this or that. It’s more, given limited resources, where can we effect the most change?
Cameron:	16:13	Yeah. I think this speaks to a growing trend where data in healthcare, which is we want to be preventative and not reactive. If we can see that there are populations that are underserved in these communities and if we just take the time to give them the basic medical care that we can afford them, for, in this case pre-eclampsia, did I pronounce that right?
Rachel:	16:31	Mm-hmm (affirmative).
Cameron:	16:33	Then, first of all, they won’t die, which is the most important thing, but they also won’t suffer. There won’t be those extra medical costs that come from when someone is sick and ends up in the hospital having to go through a situation that could affect them and follow them for the rest of their lives, however long that may be. That preventative care is obviously very beneficial.
Rachel:	16:53	Mm-hmm (affirmative).
Cameron:	16:55	I think the next most important question here is, all right, you’ve convinced me I need to get my hands on this tool. Where can we go to learn more about your work, what you’ve achieved, how we can get our hands on this tool or portal?
Rachel:	17:05	As of now, we have a couple of externally-facing publications that anyone can read [or watch]. We have presented at a few different conferences where we’re socializing the tool, figuring out how we can improve the tool. Unfortunately, the tool is not ready for public release, but we hope that in the next year or so, we can find a way to transition this to the federal government or an industry partner so that we can promote data-driven decision making to ultimately reduce maternal mortality and hopefully in turn also racial disparities in maternal health outcomes.
Cameron:	17:42	All right. Well, this has been a fantastic conversation. I’d like to say give a brief thank you to MITRE and the Knowledge-Driven Enterprise making this conversation possible, and an extremely big thank you to you, Rachel, for leading this incredibly important work. I’m so excited to see that there are these tangible results, that you’re getting to a point where people can really reap these benefits. Thank you so much for your time.
Rachel:	18:03	Thank you, Cameron.

Cameron Boozarjomehri is a Software Engineer and a member of MITRE’s Privacy Capability. His passion is exploring the applications and implications of emerging technologies and finding new ways to make those technologies accessible to the public.

MITRE’s mission-driven team is dedicated to solving problems for a safer world. Learn more about MITRE.

The United States Maternal Mortality Rate Will Continue To Increase Without Access To Data

Can Data Modeling and Analytics Help Reduce Pregnancy-Related Deaths?

An introduction to interoperability in healthcare

Marcie Zaharee and MITRE’s Open Innovation Challenge

Dan Frisk and Paula Randall on bringing innovation to government

Interview with Awais Sheikh on Deciphering Business Process Innovation

Interview with Jackie Morin on her journey from intern to senior engineer

Interview with Jay Crossler on why passion is the key to success

Interview with Dan Ward, Rachel Gregorio, and Jessica Yu on MITRE’s Innovation Toolkit

Interview with Tammy Freeman on Redefining Innovation

Is This a Wolf? Understanding Bias in Machine Learning