Data Opportunities and Challenges in a Post-Roe World
What are the barriers to conducting abortion-related research in the United States today?
Date
June 7, 2024
In 2022 the Dobbs v. Jackson Women’s Health Organization decision, which overturned Roe v. Wade, ended the Constitutional right to an abortion and dramatically changed the health care landscape in the United States. Researchers on abortion, fertility, and reproductive health have been working to understand the implications of the Supreme Court decision, including access to care, birth rates, and health outcomes.
In this webinar expert panelists discussed questions including: What are the barriers to conducting abortion-related research today? What have we learned from the data so far? Where are the data gaps and how can we fill them?
Panelists included:
- Beth Jarosz, Senior Program Director, PRB (moderator)
- Abigail R.A. Aiken, University of Texas at Austin
- Alison Gemmill, Johns Hopkins Bloomberg School of Public Health
- Laura Lindberg, Rutgers School of Public Health
- Jane Seymour, University of Wisconsin – Madison CORE
Transcript
Beth Jarosz, PRB: Hello, everyone. I’m Beth Jarosz, Senior Program Director at the Population Reference Bureau and Vice President of the Association of Public Data Users. And with both of those hats on, I want to welcome you to what will be a very engaging discussion.
As you all know, it’s been nearly two years since the Dobbs v. Jackson Women’s Health Organization decision changed the abortion health care landscape in the United States, and researchers have been working to understand the implications of Dobbs on access to care, birth rates, health, and other outcomes.
In addition to hearing a bit about that research, today we’re going to tackle questions like: What are the barriers to conduct abortion-related research? What have we learned from the data so far? What are the data gaps, and how can we fill them?
To help answer those questions, I’m joined by an all-star cast: Abigail Aiken, Associate Professor of Public Affairs at UT Austin; Jane Seymour, Research Scientist at the University of Wisconsin–Madison, Collaborative for Reproductive Equity; Alison Gemmill, Assistant Professor at Johns Hopkins Bloomberg School of Public Health; and Laura Lindberg, Professor of the Rutgers School of Public Health.
We’ll hear from all four researchers and will round out the hour with a Q&A. If you have questions, please type them into the Q&A box, and I’ll ask as many of your questions as I can during the Q&A session at the end. So without further ado, I’m going to invite Abigail to begin.
Abigail R.A. Aiken, University of Texas at Austin: Thank you very much, Beth. It’s wonderful to be here with everybody today. I’m just going to share my screen so you can see my slides. And then I’m going to give a short, 10-minute overview of some of the work that we’ve been doing trying to measure self-managed abortion and shield law abortion provision in the post-Dobbs landscape.
I want to say that the work I’ll present to you is the result of a lot of people’s work. So everybody here has been involved with Project SANA, that’s our project, the Self-managed Abortion Needs Assessment Project, at some point over the past five years. And so want to, um, say thank you to everybody who’s been involved in trying to get an effort to look at the who, what, and why of self-managed abortion in the U.S. off the ground. When we started this project back in 2018, we knew so very little in the research sense about self-managed abortion, and we really come a long way since then.
So to be clear on terms, I’m mostly going to be talking about self-managed medication abortion this afternoon. And that’s the process of obtaining medication abortion pills. It could be mifepristone and misoprostol or misoprostol alone, and managing your own abortion outside of the formal health care setting, so with no U.S. licensed provider or clinic involved. And, of course, self-managed abortion can also be done via other methods: herbs, botanicals, self-harm. There’s lots of different ways. It’s really a spectrum of things. Most of what I’ll talk about today will focus on medication self-management, but in the Q&A, I’m also happy to talk about some of the other methods as well.
So self-managed abortion is really hard to study. It’s hard to study because by definition, it’s something that’s happening in private settings. It’s usually in people’s homes. It’s not something where there’s an administrative record that you can request or track easily. And so since 2018, we have been thinking hard in Project SANA about how to, uh, count self-managed abortion, how to explore self-managed abortion, how to get a sense of how often this might occur and why it might occur.
And a lot of work has been done since then, and we really have focused a lot of our work on the nonprofit organization Aid Access. And that’s because since 2018, Aid Access has been providing self-managed medication abortion through online telemedicine. Now, the model of Aid Access has changed in recent times, and we’re going to talk about that later in the presentation. But for about five years this was considered self-managed medication abortion because it was happening entirely outside the formal U.S. health care setting.
We formed a collaboration with Aid Access, having worked with their sister organization, Women on Web, in Ireland and Northern Ireland prior to 2018. And so we were able to look at the trends in the number of people that were making requests to Aid Access. We established the safety, effectiveness, and acceptability to the user of this model. Um, please check out our papers if you’re interested in that.
And we also had a look of, at the question of, Would we expect there to be a relationship between abortion bans and self-managed abortion? You can imagine that when states put, um, abortion restrictions in place and people have less access to clinics that they might more often look to self-manage outside of the formal health care setting. And it’s also evident from our research and the work of others that people sometimes also self-manage from a point of view of being their preference. It’s not just an alternative to lack of clinical access; it’s also something people might prefer to do for a variety of reasons.
So the first part of what I’m going to talk about looks at this question of is there a relationship between SMA and abortion bans? Because leading up to Dobbs, we wanted to know, can we use data from Aid Access to find out whether when states ban abortion post-Dobbs, are we going to see an increase in people self-managing? And we had good reason to expect that that would probably be the case based on a number of prior studies that I’m going to talk about super quickly because I’ve only got 10 minutes. But I just want to show you the strength of data behind this relationship.
So this first paper looked at what happened when Texas, back at the beginning of the COVID-19 pandemic, banned abortion essentially for a period of two weeks by saying that abortion was a non-essential medical procedure. And we looked at what happened at, to request to Aid Access during that admittedly short time period. But it was a really quick and evident increase. You’re looking here at a graph of cumulative requests, so you’re seeing the actual data before the ban. They’re in black. Then you see the data after in the orange line, and the model fit, um, compared to what was forecast. And we saw this doubling of requests over the two-week time period after abortion services were shut down in Texas.
Staying in Texas, we saw what happened in September of 2021, when Senate Bill 8, that was essentially the six-week ban, uh, went into effect. And again, we saw that compared to a long baseline of what Aid Access had been doing, requests for self-managed abortion really spiked when people knew what was about to happen. But even out several months afterwards, you were seeing a tripling over the baseline numbers.
Again then after Dobbs, we saw states with total abortion bans. And again, we saw, when you compare to this baseline of what access had been seeing from these states, a doubling or sometimes even a tripling of requests in those banned states, which is more than what was happening in states without bans. So over and over again, and this is really, I think, in keeping with what we’re seeing, we see in the global context, what we’ve seen historically in other places with abortion bans, when you make clinical abortion access harder, you see this increase in people looking to self-manage.
So now what I want to talk about is going beyond requests and trying to actually count, because the question that, you know, people often ask is, okay, but we see that relationship, but how many self-managed abortions do we think are happening in the U.S. post-Dobbs? And that’s a really hard question to answer. Looking here though, this is from WeCount, and people maybe hope are familiar with the effort from Society of Family Planning to count and make a census of abortions provided within the formal health care setting after the Dobbs decision.
And this is early on, right? We’re only looking out here to six months post the Dobbs decision. But initially there was a decrease. There was a decrease in the number of abortions being provided within the formal health care setting, which raises the question of did we see a concurrent rise in people self-managing? Do we know how much of that decrease of approximately 32,000 abortions provided within the formal health care setting might get offset by abortion outside of the formal health care setting?
Now, the post-op landscape really changed what self-managed abortion looked like. That’s another issue for us. We had worked with Aid Access for a long time and continued to do that because they were an online clinic mailing pills. But there was such a response to the Dobbs decision from community support networks. So based on accompaniment models, oftentimes in Mexico or Latin America, and we knew that model’s been so prevalent in South America for so long, coming to the U.S. to try to secure access for people through volunteer networks and then also websites selling pills. So not online, um, telemedicine operations, but simply online vendors that were like, yep, we’ve got misoprostol/mifepristone. You can order it from us, and we’ll send it to your house.
So trying to, first of all, map this is a large effort, right? Trying to figure out how many different providers out there, out there, as particularly with online vendors, they tend to change a lot. And different people may run multiple websites. They may pop up and go away. It’s hard to keep track of all this. So part of the work that we’ve been doing is trying to figure out how many pathways are there, how many providers are there, and then what does each provider provide.
And in an attempt to do that, we published the first kind of findings out of that looking out six months post-Dobbs. Um, you can find this paper also online. We saw this increase in self-managed abortions provided in that six-month period post-Dobbs, when you saw the decrease that was found within the formal health care setting and now this increase in self-managed abortion. And this chart attempts to break it down by the different types of provider, the community networks, the telemedicine org, and the online vendors. So you can see the baseline.
There was, of course, provision beforehand before Dobbs. People were, of course, self-managing then, too, partly as a result of abortion restrictions, partly as a result of preference. But it really changed post-Dobbs in mostly in states that enacted bans. And approximately 26,500 abortions is our estimate for the six-month post-Dobbs that took place outside the formal health care setting. Lots more methodological details on that in the paper that I don’t have time to cover right now.
I wanted to highlight that these different types of organization, um, play a critical role, partly because of cost. Um, from our qualitative work, we know that even the $90 that a service like Aid Access was asking for in donation is too much for a lot of people, and many of the accompaniment networks are able to provide at no cost, which is really important for people.
Secondly, these options are out there. They are, they’re, they’re becoming more known about. But that doesn’t mean that people don’t often look to other ways, too, maybe as part of their journey to getting what they need or as the ultimate end of their journey. And for some people, this could even be something that’s unsafe or harmful to them. So I haven’t talked a lot about that today, but I don’t want to overlook it completely. It comes up in our qualitative work quite a lot.
Then finally we’ve got this shifting post-Dobbs landscape. We’re now looking at numbers coming in from WeCount showing abortion numbers increasing. And we also see from the Guttmacher map project also, um, similar counts in the monthly provision data as well. So there seems to be an increase in access overall.
And what Aid Access are now doing we wouldn’t call self-managed abortion anymore because they’re now providing, um, entirely through U.S.-based clinicians, through shield laws. So getting medication abortion to banned states from providers that are working in states that have shield laws, that allow that to be within their scope of practice. So now we have a different, um, challenge on top of the counting of self-managed abortion, which is this new definition of shield law provision, and trying to count that alongside, um, the self-managed, uh, medication abortions too.
So I like to leave it there because I only have those 10 minutes. Uh, that’s a summary, a very high-level summary of what we’ve been up to and what we’ve been trying to do. Uh, please check out our site to look at our papers. Um, I’m looking forward to your questions in the Q&A as well. Thanks so much for your time.
Beth Jarosz: Thank you so much for setting the stage and for covering a lot of information in a very short time. And I’m going to invite Jane up next.
Jane Seymour, University of Wisconsin–Madison CORE: Wonderful. Thank you all so much. Um, I’m really delighted to be here today to talk about some of the ways that we’re measuring the impact of Dobbs in Wisconsin at the University of Wisconsin Collaborative for Reproductive Equity, or UW CORE. CORE is a research initiative focused on abortion, contraception, and other aspects of reproductive autonomy that’s housed at the University of Wisconsin School of Medicine and Public Health, which is Wisconsin’s largest and only public medical school.
Okay, there we go. So given our focus, one of CORE’s goals both pre- and post-Dobbs is to document the health, well-being, and social consequences to Wisconsinites of barriers to wanted abortion care.
As you may know, post-Dobbs, an 1849 state law here in Wisconsin was interpreted as banning abortion, and as a result, all abortion services in the state were halted, and data from the Society of Family Planning’s WeCount effort, which Abigail referenced, shows here that the, in the year after Dobbs, roughly 7,000 fewer abortions occurred in Wisconsin compared to the prior year.
And while abortion services have resumed in Wisconsin as of last fall, we know that many barriers that predate Dobbs remain in place, and providers are still ramping up services to pre-Dobbs levels. In other words, there are still significant barriers to abortion in Wisconsin.
So naturally, this leads us to ask what happens to Wisconsinites who want abortion services given these bans and restrictions. And we can imagine three scenarios for these folks. First, we can imagine that some likely cross state lines for abortion, as was already the case prior to Dobbs, given Wisconsin’s extreme abortion restrictions. However, data from, excuse me, data from WeCount, um, shows us that the increases in haven states, such as some of the other Midwestern states shown here, don’t make up for the bans in states such as Wisconsin. Um, second, we can imagine that some people self-managed an abortion, obtaining pills from a variety of sources, including places like Aid Access. And finally, we imagine that some Wisconsinites who wanted an abortion did not have one and instead carried their pregnancy to term.
So although we can imagine these scenarios, it’s important to understand the lived experiences and trajectories of abortion seekers in this restrictive environment as well as the impact on their health and well-being.
So to understand these experiences, we must get information from Wisconsinites who considered abortion. As many on this call know, prior to Dobbs our field often recruited for studies from abortion clinics, which may have induced a selection bias, as we failed to include those who face barriers to care such that they never made it to a clinic. Some researchers in our field have made attempts to overcome this limitation by recruiting from prenatal care clinics and/or online via social media or, or Google ads.
Post-Dobbs, especially in states where there were no longer abortion clinics, including Wisconsin for a time, this method, methodological challenge has only been amplified. In other words, we have to search in many different places for research participants to fully answer our questions about barriers to care.
Here at CORE, we’re taking a multi-pronged approach to understanding the impact of Dobbs on abortion seekers, which we call the post-Roe impact research portfolio. In this portfolio, there are three studies shown here, which we refer to as the Turnaway, prenatal, and MAP studies. I’ll briefly note that the Turnaway work is comprised of interviews with those who participate in Dr. Diana Greene Foster’s post-Roe, uh, work.
But today I’m going to focus on our other two studies, which recruit from prenatal care clinics and online, respectively. And both include longitudinal surveys as well as in-depth interviews. Overall, this portfolio builds on Dr. Foster’s original Turnaway Study as well as pre-Dobbs work that sought to recruit those who considered abortion outside of the clinical setting.
Here’s a bit more detail about these two studies. In brief, the prenatal study recruits patients from UW Health prenatal care clinics, after which they complete a baseline survey. Those who report having considered abortion are invited to participate in an in-depth interview two weeks after the baseline survey, as well as for 10- and 18-month follow-up surveys. Additionally, at baseline, we ask all participants to agree to use of their electronic medical record, or EMR. We pull EMR data for all study participants to compare outcomes between those who did and did not consider abortion care.
Oops. Excuse me. Uh, currently, we’re still recruiting, and the first participants have received their 10-month follow-up survey. As you’ve already seen, the MAP Study, or the Midwest Abortion Pathways Study, is a partnership between CORE, Ibis Reproductive Health, and Indiana University and recruits participants via Google and Microsoft internet search engine advertisements. Participants are eligible if they’re pregnant, live in Wisconsin, and report having considered abortion for their current pregnancy. They complete a baseline survey after clicking through the advertisement and then are invited to complete 4-, 10-, and 18-month follow-up surveys, as well as an interview post-4-month survey. Currently, we’re still recruiting in the first. Participants are about to receive the 10-month survey.
Now, typically this is where I would share a few nuggets of our results, but instead I’m going to share some challenges we’ve encountered as those feel particularly relevant to today’s conversation.
First, as is likely no surprise, abortion seekers who were hard to find pre-Dobbs are even harder to find now, likely due in part to concerns about the legality of abortion and related increases in conversations about digital security. Second, the legal and health care delivery context for abortion is extremely dynamic. We’ve had to be very flexible and in some cases act quickly to update our survey instruments to understand the current legal or political context in Wisconsin.
Finally, and perhaps most notably, bad actors and bots abound in this work. We’ve been overwhelmed by the number of fraudulent responses we’ve received and, in turn, the time and effort it takes to parse the good responses from the bad. In the cases of bots that churn out thousands of responses each week, identification is fairly easy, as, for example, their open-ended responses and their email addresses are completely nonsensical. But for those duplicate responses from legitimate participants who are likely trying to obtain additional remuneration and individuals posing as eligible when they’re not, the process is much more art than science.
Through this process we’ve learned a lot about how to proactively prevent fraud and handle it when it slips past our checks. First, the thoughtful, capable data scientists on our team have been an invaluable resource. Their efforts have allowed us to analyze data quickly and again, pretty proactively and quickly, identify appropriate data flags with relative ease and pivot our processes accordingly.
Additionally, having checks in our data collection tools that help identify bad actors and bots has been crucial to our success. I’m happy to talk more about those in detail during the question and answer period, but for now, I’ll highlight to you. Um collecting IP addresses and asking questions for which responses can be compared, both within a single survey and across multiple survey waves has been crucial in helping us catch fraud. With that said, I’ll say that our team remains hopeful in this really is a team effort.
Through the process, we’ve developed even stronger partnerships, and we’ve learned so much. We’re better scientists because of the challenges we’ve encountered. And while our progress may be slower than we hoped in some regards and much more challenging than we expected, our methods are working. We’re finding Wisconsinites who considered abortion and never made it to a clinic and are willing to share their experiences with our team.
So I’ll leave, leave it at that except for two asks. My first is that we continue to fill the gap in finding folks who consider abortion but don’t make it to care. They’re a crucial population within, uh, that should be a focus of our work more often, and I’d ask that we all continue to work collaboratively to build effective anti-fraud strategies that can be applied across contexts. I know there are many folks in the field and in other fields working on this effort, and our collaborative efforts have already been so fruitful, and I welcome more investment.
So with that, thank you very much. And I’ll welcome questions during the question and answer or at this email address at any time.
Beth Jarosz: Thank you so much, Jane. And we will move on to Alison.
Alison Gemmill, Johns Hopkins Bloomberg School of Public Health: Okay. Great. So thank you so much for organizing this. Um, I’m really happy to talk about the, see, it’s working, the data component here. So, um, for those of you who don’t know me, I’m a demographer and a reproductive perinatal epidemiologist, and I am studying the health impacts. And I’m going to talk about some of the challenges of using secondary data sources in this work.
Um, so I wanted to highlight two current projects. Uh, and this is collaborative work with Dr. Suzanne Bell. The first of these is to look at the impact of highly restrictive abortion policies on fertility: so birth rates, uh, severe maternal morbidity and mortality, if possible, and birth outcomes. And then the second project is to look at the impact of these highly restrictive abortion policies on changes in high-risk pregnancy care. And this work is currently funded through the Society of Family Planning.
So what are the data sources that we use? Um, for the first project where we look at birth rates, we obviously use birth certificate data. So we, we rely on vital statistics data a lot. Um, we are also looking at some infant outcomes like infant mortality. So we rely on death certificates as well. The good thing about vital statistics data is that they are virtually complete. So for births, for example, they represent over 99% of all births in the U.S., including home births, and all states participate. So we have good coverage over time and space.
Um, for maternal health outcomes or pregnancy-related outcomes, we can’t really rely on vital statistics for those. And I’ll explain more why shortly. So we’re going to rely on state-level hospitalization data, um, and specifically this is through a database through AHRQ. And what we do is we use ICD 9 or 10 codes for diagnosis and procedure, procedures to identify these types of outcomes.
And then, uh, potential projects in the pipeline might actually be using some of our typical survey data. So, for example, some of you are familiar with the National Survey of Family Growth, or NSFG, and the Pregnancy Risk Assessment Monitoring Survey, or PRAMS, and I’ll talk about those in a bit.
So I wanted to actually highlight some data challenges, uh, because I think it’s really important in this work, and it’s what we’re dealing with. The number one challenge: everybody wants to know what’s happening on the ground right now, but we do not yet know, and that’s because our data are lagged in terms of when they are released. So in the case of vital statistics data here, um, usually final birth and death certificate data are lagged by a year. So we have to play the waiting game. However, uh, I’ll talk about this at the end, uh, there have been changes in that they’re making provisional data more available to researchers. And this has been an amazing change.
Um, in terms of hospitalization data, these take a lot of time to come out as well. So I can’t tell you yet what’s happening to an outcome like severe maternal morbidity. Um, so these are lagged by about two years. So data on 2023 births will not be available until early 2025, for example. Um, in terms of those two surveys I mentioned, CDC’s PRAMS, they interview people who are postpartum, so these are among people who gave birth, data on 2023 births will not be available until 2025. And then for the National Survey of Family Growth, they did resume data collection in 2022. So there is a potentially an opportunity to use these data, but they won’t be released for some time.
The second challenge is about varying data availability, availability across states or limited geographic identifiers, and to conduct the kinds of studies that we want to conduct that have very strong causal inference embedded in them. We need data for states, each state. So the state inpatient databases that I talked about, the hospitalization data, those are great because we can get those at the state level. However, one of the big states that we’re interested in, Texas, does not participate, at least in the HCUP Central Distributor where we access these data typically. And then for the NSFG, they do not include state identifiers, nor do they include month year of important reproductive events. To access those data, you actually have to go to a restricted data center, which is associated with time and cost burdens.
The third challenge I want to note is that there’s varying data quality across states. And this is very, or this is, um, specific to birth certificates here. Um, so not all items on the birth certificate have high validity. And this validity varies by state. So because I have a captive audience, I wanted to highlight one of the papers that we wrote. Um, so what you see down here is a section of the birth certificate where there is something known as the maternal morbidity checkbox. So on the birth certificate we can capture information about some of these common maternal morbidities like maternal transfusion or ruptured uterus. However, we did a validation study where we compared the data on the birth certificate to the hospitalization data, and the estimates do not match up. And what our conclusion was is that the birth certificate data for these specific maternal mortality, morbidity indicators, the validity is quite low. So I recommend not using them.
Another challenge: the data are cost prohibitive. So I mentioned that we’re going to be using these state inpatient databases. These are really expensive. So one year of data for one state can cost anywhere from $200 in the case of Florida to about $1,600 in the case of Mississippi. So for our project where we need data for several states and several years, this is going to be quite costly. Thankfully, we have funding to support this work. Um, but again, this is a big barrier, right? Um, I mentioned the NSFG. You have to go to that restricted data center, and that costs a lot of time and a lot of money.
And then finally, I think a really great data source are Medicaid claims data. However, I will not be using them because they are very expensive to use. And you kind of need an existing data infrastructure. So you need, so you need to be linked to universities that are already kind of using Medicaid data because it’s a big, um, what’s the word, it’s a big barrier to overcome to get started with the Medicaid data, is what my understanding.
And then finally, this is my last data challenge, and it’s more of a statistical challenge. It’s about how do we estimate impacts among subgroups. And we know this is a very important question. However, the numbers can be quite small and this comes with lower statistical power. So, for example, how do we measure fertility rates among smaller or, yes, subgroups in smaller geographies? Or how do we measure events of very rare maternal outcomes that really matter, but they might just be so rare they’re hard to study.
So we need to expand our causal inference toolkit to detect these effects. We don’t want them, we don’t want them to go unnoticed, right? We want to detect some signal. So how do we do that? And that’s where I think a lot of the work needs to be done in terms of population health outcomes.
And then finally I’ll close with some data opportunities. Um, the one that I briefly touched on is that the CD, or the National Center of Health Statistics has made this release of provisional data in terms of birth certificate data and death certificate data. Um, and it’s been a game changer. I think what happened is COVID prompted the release of this data because we needed to know in real time, especially with mortality, what was going on. Um, and as of 2023, provisional birth and death certificate data are now available on CDC Wonder. So if you don’t know about CDC Wonder, this is a great interactive, um, platform that you could use online to download data. It’s an amazing resource.
So one of the first things that we did, um, while we’re still waiting for, uh, to study impacts of Dobbs, is we could study the impact of Texas SB8 policy using that provisional data from NCHS. And so this is a paper that my colleagues and I wrote, um, looking at what happened to fertility after SB8. And we did find a 3% increase in live births.
And I believe this is the last thing I want to say. So I just wanted to know other potential data sources for those of you in the audience that are thinking about doing this work. Uh, one colleague at Hopkins, he’s a current student, has actually used the behavioral risk factor surveillance system, which does have data by state. And he looked at outcomes related to self-reported anxiety and depression and found that those were elevated in respondents in banned states following the Dobbs decision.
I know other people might be looking at changes in the workforce. So there’s potential data, um, from some organizations that might be possible. I know people have looked at Reddit forums. Um, so there’s some rich qualitative there, data there potentially. And then I think we’re just going to be, have to be innovative in terms of other types of digital data that might be used.
So, um, I think that was it. And I look forward to any questions you might have. Thank you so much.
Beth Jarosz: Thank you so much, Alison. And last but certainly not least, I will invite Laura to present.
Laura Lindberg, Rutgers School of Public Health: Okay. Thank you all for having me here today. And, um, Alison really set this up because I want to focus on a subpopulation, which is adolescence. And I want to move my slides. There we go.
So adolescents are experiencing disproportionate legal, financial, logistical, and social barriers to abortion. This policy environment impacts not just adolescents seeking abortion, but all adolescents. And the experience of adolescence itself may have fundamentally changed. Adolescents are thinking differently about many aspects of their lives: their relationships, health, where to go to college, where to live, and what their future might look like.
Adolescents are impacted by new abortion laws aimed at all pregnancy capable people, as well as those such as abortion trafficking laws that specifically target minors’ access to abortion. And abortion access remains difficult for minors, even in states where access is protected, whether it’s because of parental involvement requirements, financial logistical challenges, or forms of abortion stigma.
It’s against this backdrop that I’m going to draw on a new report, Adolescence Post-Dobbs: A Policy-Driven Research Agenda for Minor Adolescence and Abortion. And I’d like to take this opportunity to acknowledge and thank my incredible co-authors, Julie Maslowsky and Emily Mann. While we focus on minors because of their unique standing in the law, our recommended action steps would benefit adolescents of all ages, their families, and their communities. And today, I’m going to focus on the data agenda that we developed in this report.
This report was produced under the auspices of Youth Reproductive Equity, a national multidisciplinary research collaborative composed of both researchers and clinician scientists. We formed in 2021 in anticipation of the Dobbs decision and its disproportionate impact on young people, and we continue to expand our work.
So, we found that minors are systematically underrepresented in research about abortion. Far too often, studies start at age 18, excluding the experiences of younger adolescents. Failing to produce needed research on minor adolescents and abortion is an equity issue, and the large knowledge gap has become a liability as the data gap allows for non-evidence-based policies.
A key call to action of our research agenda is to ensure that research designs and analyses include the experiences of minors. We make a distinction between studying the direct impacts of changing abortion policies on pregnant minors and the indirect impacts of abortion policies on the total population of minors.
So currently we lack the data needed to study the direct impacts of the new restricted abortion policies on minors. For example, it’s well established that federal and state abortion surveillance is incomplete. For example, California doesn’t participate in these systems, and states don’t always collect data by age. Studies of abortion patients, usually based out of clinics, even when they do include minors, are often limited by small numbers, and the new real-time data collection of abortion counts, such as that from Guttmacher or the Society of Family Planning, doesn’t even collect patient age, leaving critical gaps in the surveillance of minors’ receipt of abortion care.
Thus, we recommend expanding data collection to increase and improve the inclusion of minors in clinical studies, as well as state and national surveillance, and this may include a need targeted oversampling of minors.
Further, where there is data, we need to expand again our approach so that we present age-specific data in ways that we can identify minors’ unique experiences and not group them with all adolescents up to age 19 or, worse, with the general population.
In addition to expanding our research, we call for approaches that use tailoring, which is to tailor direct collection to provide an in-depth examination of those experiences that are unique to minors. This allows for focused attention to policies, focus on this age group’s abortion access, experience, access and experiences separate from those of adults. And a key recommendation around tailoring is to field a new longitudinal study of pregnant minors across different policy environments to better understand their pregnancy, abortion, and parenting experiences.
Now, I want to turn to the data needed to examine the indirect effect of abortion policies on minors. And here we propose the need for what we call contextualizing, calling for population representative as well as targeted studies of minors that aren’t so focused on abortion but capture the context of adolescents’ lives as abortion access is changing.
So as we think about contextualizing, we can see many gaps in existing federal, state, and national data collection efforts that limit our, our ability to do needed research. So let me just talk about a few. There’s obviously other data sources out there, but I think these are some major ones that are worthy of discussion.
So the Youth Risk Behavior Survey, or YRBS: these are state surveys of high school students. And they should allow us to compare between different abortion policy environments. But an increasing number of states are choosing not to participate in the YRBS, and this is likely to only get worse over time. Still, there are opportunities here to abortion policies by knowing the state that the student resides in to outcomes such as their mental health, their experience of intimate partner violence, and contraceptive use patterns.
Alison mentioned the National Survey of Family Growth. This is a household survey that starts at age 15 and goes through age 49. However, the sample size of adolescence is relatively small, and especially if you want to do analyses limited to sexually experienced teens. Um, Alison noted that the geographic identifiers in the study are not made publicly available, which I’ll talk about more in a minute. She also mentioned BRFSS, and this is a good resource because it does provide state representative health data that could be of interest, but it only samples adults. And this is an example of the exclusion of minors from research that an expansion of the survey could address and improve.
And finally, we lack a current longitudinal study of adolescent lives. Add Health has been probably the most influential source, source of longitudinal data on teens, but it was started in the 1990s, and those adolescents are now adults. Indeed, it’s the National Institute of Aging that now funds this project, so it doesn’t help us to study today’s teens in today’s post-Dobbs world.
The National Longitudinal Survey of Youth, or NLSY, faces similar aging as the 1997 cohort, which was the most recent cohort, is now in middle age. The Bureau of Labor Statistics is currently designed designing a new NLSY, and this is really an opportunity, I hope, for collaboration to ensure that relevant health and psychosocial and other effects are, and measures are included in addition to the conventional labor force and work and education measures that this survey has usually focused on.
These gaps lead to key recommendations to improve and expand current data collection and start new efforts in the field. So we’re calling for both new cross-sectional and longitudinal survey of the general population of adolescents. These surveys should include not only sexual and reproductive health behavior, but adolescents’ knowledge, attitudes, and behaviors related to changing abortion access. And they should allow for the longitudinal study of the impact of the Dobbs environment of living in this, at this time on their behaviors, their education, their economic and their health, health outcomes into adulthood. Now is the time to design and implement these studies.
And we recognize that part of the context of minors’ lives are the adults in them. Be it parents, health care providers, caseworkers, teachers, even policymakers whose views on adolescence color their approaches. And studies of these adolescents who are influential in minors’ abortion experiences are also needed to understand the context of these experiences.
Finally, I want to remind everyone that in our country right now, your zip code determines your access to abortion care. And to help researchers study the influence of location, we need to make geographic data more readily accessible. This could include strategically collecting state representative data as well as making existing geocodes on surveys more available. So Alan, Alison pointed out the challenges of the difficult to access NSFG geocodes through the research data centers. I’ve lived through that; I call it often the, um, circles of hell, um, and it is not easy. But one approach to facilitating needed research with these geocodes could be to create publicly available aggregated geocodes that group states according to their state policy environment but don’t run these kinds of risk of disclosure, disclosure that the RDC is trying to protect from.
So in conclusion, all adolescents are impacted by the changing abortion policies, even if they aren’t seeking an abortion. And there’s a need, there’s substantial need for more research and data for this population. The historic exclusion or blind eye to minors’ experiences as compared to adults leaves us with inadequate data systems. Excluding and overlooking minors is both an equity and a rigor issue. We need quality science that includes marginalized populations, including those treated differently because of their age. Expanding, tailoring, and contextualizing data collection for minors and improving how researchers can access key data offers us a needed path forward.
I’ve shared here a QR code so you can access, download our entire report. It has not just these data and research needs, but also a deep dive into the changing legal and policy environment. I also invite you to contact us at Youth Repro. We are available for consultation, collaboration, and thought partnership, and my email is up here as well. Thank you.
Beth Jarosz: Thank you so much. All of these presentations have been fantastic, and I want to take a moment to acknowledge all of the great work that you are all doing, and also to suggest that our audience members seem to think so, too, because we have a ton of questions. Um, I will try to get to as many of these as we can, and that’s the 15 minutes that we have left.
And so I want to start with, um, and this is probably for Abigail, but for any of these, any of you can answer, um, and the question is, Is it legal for women in banned states to receive the pill in the mail? And I think maybe talk a little bit about shield laws, which you mentioned in your talk.
Abigail R.A. Aiken: Yes. Thanks for the question. Um, it’s a complicated one because it depends, um, on who we’re talking about being the subject of the laws. Um, it is the case that, um, most states don’t have laws on the books currently that would explicitly criminalize the person using the medications or receiving the medications for a self-managed abortion, although that doesn’t mean that people won’t be surveilled and won’t be subject to investigation or even times prosecution, um, unlawfully. And so, um. That’s one where I would also, if you’re interested in that question, check out the resources of If/When/How: Lawyering for Reproductive Justice because they’re extremely knowledgeable on this issue. So that’s one, um, where it’s, you know, not explicitly criminalized, but doesn’t mean that people couldn’t face, um, legal jeopardy.
For those who are delivering the pills, and I see the question about, um, is it, uh, legal for the person sending them. Now, technically, that would be against the laws of most states who had to have abortion bans or have restrictions on, uh, telemedicine provision of abortion. But the idea of the shield laws is to protect providers in states where they reside and where they practice. So there’s a great article in The New England Journal, uh, written by David Cohen, that lays out shield laws. It’s a really interesting and informative read, and it tells you about some of the protections, uh, that providers residing in states with shield laws would have in terms of protecting their license and protecting them from states that want to enforce their own state laws outside of their state boundaries.
Beth Jarosz: You know, and that kind of leads into, I love someone posted a question that I already had on our list, and I think it dovetails nicely with this. And it’s speaking to privacy and confidentiality. I think that’s one of, it’s sort of the, the elephant in the room when we’re thinking about this, that we need really good, high-quality data for a topic that is sort of legally challenging. Um, and, uh, so can you speak, I know we talked a little bit about that with geography, but I think each of you probably have a perspective on this. Who wants to go first, Abigail, do you want to take us off?
Abigail R.A. Aiken: Yeah. I can kick off there. Yeah. From the self-managed abortion and also the shield law perspective, um, it’s extremely difficult, right. And, you know, we know that, as I said, just because people don’t live in states with state laws that explicitly criminalize them doesn’t mean they won’t be, uh, surveilled and harassed and sometimes even unjustly prosecuted.
And so we really limit our data collection, and we are really limited in terms of what we can collect. So we never collect anything identifying. And even then, we don’t collect a lot of the things that, you know, I appreciated Laura’s call for this more detailed data, and I think we absolutely need that. It’s really hard with self-managed abortion. Um, and so far we have really stuck to the idea of firstly, anyway, counting right, getting information on volumes on prevalence.
Um, before Dobbs, we did quite a lot of qualitative work looking at people’s experiences and their motivations. And I’m not saying we won’t go there again. Um, but it is an even more difficult environment in which to, uh, do this work than it was before. And so, um, right now we’re really very much, when we get data directly from providers, we ask for as little as possible.
Beth Jarosz: Thank you. And I think Alison and Laurie, you both mentioned, um, sort of geographic specificity, which we know is really important in this context. And, Jane, I have a follow-up related to privacy for you, too, but I want to talk a little bit about the challenge of balancing geographic access with confidentiality in these cases. If either of you want to speak to that.
Alison Gemmill: Well, I was actually thinking about some of the rarer outcomes that we’re going to be studying, which you could inadvertently disclose somebody’s identity in a given state, you know, in a given age range. And they have a very rare pregnancy-related outcome. And so one of the challenges we’re going to have is that data are made available to us, and we have a data use agreement where we will make sure that we protect the data at all, at all costs.
But, um, it’s, it might be challenging for us in how we disseminate the information. So we have to make sure that we’re not going to, you know, report only five cases of something. So there’s certain rules. So that’s how I’m thinking about it with the secondary data that I use.
Laura Lindberg: Yeah. And I just want to mention, I mean, there certainly are real risks here. And our role as researchers, we have to take those seriously. But we also need to be educating our IRBs about the reality of the extent of those risks. And what we’re hearing from a lot of research these days is that IRB, IRB members don’t know much about abortion. They’re getting their news from wherever they’re getting it. Um, and they may be concerned in ways that doesn’t reflect the true risk, and their solutions may not be true solutions.
So being the person in your university or in your setting who can work with the IRB to educate them, um, can be helpful. And the Society of Family Planning is in the process of preparing and will be disseminating a series of, um, guidance documents that people can use with their IRBs, both for general sexual and reproductive health research and focused on doing research with minors. So that should be useful to the field.
Beth Jarosz: Wonderful. Thank you. And that’s actually a really nice lead into the question that came in for Jane. And it was thinking of privacy, actually, from the flip side is that, um, you’re finding ways to remove cases that are fraudulent using IP addresses, but does that cause any IRB or confidentiality concerns?
Jane Seymour: Thanks so much for this question. This is something I think about pretty much constantly, it feels like. Um, I think it’s a really, really important conversation, and the tension is really real. So I don’t, I don’t pretend to have any of the answers.
Um, with that said, I really appreciate Laura’s call to, for high-quality data. And we as researchers have a responsibility to ensure that the data that we put out into the world, the results that we put into the world are as real as possible. And in the case where we’re dealing with hundreds and thousands of cases of people who in some cases, like, I think they’re, they’re kind of the two groups I spoke about, there’s like the bots and things that are really, really easy to weed out. You know, when it’s like Abcdefg and a string of 14 letters at mymail.com. That’s pretty easy.
There are some people who have, like, really done their homework and have really convincing stories. And in some cases, we’ve gotten as far as getting them on the phone for an in-depth interview, and it’s become clear that there’s no way that this is a real story that they’re telling. They’re talking about getting pills over the counter in Wisconsin in a time period when, like, pills are not available over the counter, abortion wasn’t available in Wisconsin, you know, just and, and when we probe that, it’s clear that that’s not the case. We’re not talking about issues of stigma where somebody might be changing their story due to abortion stigma.
At any rate, I think it’s incumbent upon us to balance participant safety and security with the rigor and validity of our data. And we feel pretty strongly that IP address is one of those data points we can collect that helps us significantly. We’re really lucky at our institution to work with an IRB that’s very supportive of our work.
One thing I’ll flag here that I didn’t have time to talk about in my presentation was the challenges that we’re working with, um, as it relates to certificates of confidentiality from the National Institutes of Health. There have been changes to that program, and many third-party platforms, including Qualtrics, which I imagine many of us use for data collection, are no longer acceptable, uh, third-party platforms to use under a certificate of confidentiality. So we’re really again struggling with how much does the rigor, how do we balance the rigor with the data protection. And again, I haven’t figured this out, but I think it is a real tension that we’re going to have to continue to work with as a field.
Beth Jarosz: So, um, so, so many more questions. We have time for one, maybe two more. So let’s see if I can package these together. Um, I have sort of a very broad question. I think all of you have touched on this, and that is, What’s a research question you really wish you could answer but don’t have the data now? And one specific question from an audience member is, Do you have statistics on covert delivery? Um, sort of these, uh, abortions that are happening in the states that, that perhaps that have banned abortion, um, in the wake of Dobbs.
Alison Gemmill: Um, I guess I’ll start. And mine is a pretty easy ask. I think I talked about state inpatient data, so that’s people who are hospitalized. I think the next step would be thinking about emergency room departments as a source of care. Um, and those data do exist, but again, not for every state. So that’s a big challenge. And then I think, like what Laura said about longitudinal data, I would love to have longitudinal data to link people over time, whether that’s within administrative claims data or a survey using secondary data. So I, I would love to see data like that eventually to answer some of the questions that we have.
Beth Jarosz: Laura, I know you had a call for data on adolescents. Do you want to renew or repeat them?
Laura Lindberg: Yeah. I mean, I think for, for adolescents, from a research perspective, from an IRB perspective, from a policy perspective, we need to be thinking more about what the harms are for adolescents who wanted an abortion and couldn’t get it, who want and were not telling their stories in our research, that the harm that when we think we’re protecting them by not including them in our research. In fact, so many harms happen when we can’t tell the stories and they’re not included.
And this happens not just for abortion research, by the way, but sexual reproductive health research more generally. So do we, if we don’t include minors in our contraceptive studies, we can’t show that contraception is safe for them because they weren’t in the study to begin with. They need to be included more to in fact increase their safety, not harm it.
Beth Jarosz: And I think I have one more question for Alison. And it’s, you know, you talked about the maternal morbidity data and the research that you’ve done around the challenges with that on the birth certificate record. Um, if you could make recommendations for how to improve, and you probably have that in your paper, but if you could make recommendations about how to improve that data collection system, what would you say?
Alison Gemmill: Yeah. I think, thank you. That’s great. And it was a research letter so I couldn’t say much, but I would, I mean, my understanding is that the National Center for Health Statistics has not had the resources to really check the validity of items. Um, and so first of all, I would want an evaluation of the validity of items on the birth certificate. That would be number one. But then second, it’s really a state’s issue, right? And the data are collected within states at their health departments. And it seems like there is varying data quality. So if there’s a way to train people that fill out the birth certificate to improve the way that they do that, I think that would go really far, because there’s some pretty potential rich data on maternal health there that we could be using.
Beth Jarosz: I think we have time for one more question, and I think this one is for Jane. And it’s, How does a researcher assure that a source for qualitative data like Reddit is reliable and meaningful?
Jane Seymour: That’s a fabulous question. Um, I think that in qualitative data, you know, the aim is not to be generalizable. The aim is to understand the experiences of those people for whom data exists. Um, you know, unfortunately, this is, I have not worked, some of my colleagues at CORE have worked with Reddit. Um, I have not worked directly, worked with Reddit data, but know that there’s quite a literature, which Alison you’ve referenced here. And Alison, forgive me, I’m not sure if you’ve worked with Reddit data, so please feel free to chime in if you have, or anyone else on the panel.
But I think that that Reddit data is a source of data that we can use to understand the experiences of some people who are searching for abortion, have had abortion, or have been denied abortion. And it’s never going to give us the full picture of the experience of everyone, but I want to be really clear that a lot of our quantitative research also doesn’t do that. So these are all important pieces to a larger puzzle that we can put together when we work effectively together to, to gather high-quality, rigorous data.
Beth Jarosz: Thank you. So I want to thank you again for sharing your expertise and your time with us today.