糖心vlog

Key Points

Question听 To what extent has the provision of medications for self-managed abortion outside the formal US health care setting increased in the 6 months following the Dobbs v Jackson Women鈥檚 Health Organization decision?

Findings听 In this cross-sectional US-wide study, the total number of provisions of medications for self-managed abortion increased by an estimated 27鈥�838 in the 6 months after Dobbs (July-December 2022) vs what would have been expected based on pre-Dobbs levels.

Meaning听 Provision of medications for self-managed abortion in the US increased in the 6 months after Dobbs. A substantial number of people accessed abortion medications despite implementation of state-level bans and restrictions. Clinicians should be prepared to meet the needs of patients who may be considering self-managed medication abortion or who need or want to connect with the formal health care setting for postabortion care.

Importance听 The Supreme Court decision in Dobbs v Jackson Women鈥檚 Health Organization overturned the right to choose abortion in the US, with at least 16 states subsequently implementing abortion bans or 6-week gestational limits. Prior research indicates that in the 6 months following Dobbs, approximately 32鈥�360 fewer abortions were provided within the US formal health care setting. However, trends in the provision of medications for self-managed abortion outside the formal health care setting have not been studied.

Objective听 To determine whether the provision of medications for self-managed abortion outside the formal health care setting increased in the 6 months after Dobbs.

Design, Setting, and Participants听 Cross-sectional study using data from sources that provided abortion medications outside the formal health care setting to people in the US between March 1 and December 31, 2022, including online telemedicine organizations, community networks, and online vendors. Using a hierarchical bayesian model, we imputed missing values from sources not providing data. We estimated the change in provision of medications for self-managed abortion after the Dobbs decision. We then estimated actual use of these medications by accounting for the possibility that not all provided medications are used by recipients.

Exposure听 Abortion restrictions following the Dobbs decision.

Main Outcomes and Measures听 Provision and use of medications for a self-managed abortion.

Results听 In the 6-month post-Dobbs period (July 1 to December 31, 2022), the total number of provisions of medications for self-managed abortion increased by 27鈥�838 (95% credible interval [CrI], 26鈥�374-29鈥�175) vs what would have been expected based on pre-Dobbs levels. Excluding imputed data changes the results only slightly (27鈥�145; 95% CrI, 25鈥�747-28鈥�246). Accounting for nonuse of medications, actual self-managed medication abortions increased by an estimated 26鈥�055 (95% CrI, 24鈥�739-27鈥�245) vs what would have been expected had the Dobbs decision not occurred.

Conclusions and Relevance听 Provision of medications for self-managed abortions increased in the 6 months following the Dobbs decision. Results suggest that a substantial number of abortion seekers accessed services despite the implementation of state-level bans and restrictions.

Since the US Supreme Court decision in Dobbs v Jackson Women鈥檚 Health Organization, 14 states have implemented total or near-total abortion bans and 2 have implemented 6-week bans.¹ These bans have greatly reduced access to abortion across large geographical areas, particularly in the southern US.² Although people may still travel to access care, this option is unaffordable or logistically impossible for many.³ The immediate result was a decline in the number of abortions provided within the formal health care system: across all US states, an estimated 32鈥�260 fewer such abortions took place in the 6 months after Dobbs (July-December 2022).⁴

The objective of this study was to answer the natural follow-up question: Over that same 6-month post-Dobbs period, how many more people obtained abortion medications from sources outside the formal US health care system, and how many individuals subsequently used the medications to conduct a self-managed medication abortion? Self-management can include a spectrum of methods, including herbs, botanicals, and self-harm.⁵ Increasingly, self-management involves the use of abortion medications, whether mifepristone and misoprostol together or misoprostol alone.⁶ Before Dobbs, people in the US were already self-managing in this way, particularly in states where in-clinic abortions were heavily restricted.⁷ Prior research has found increased demand for self-managed medication abortion in response to sudden policy changes, such as short-term bans on abortion under COVID-19 restrictions in 2020 and the implementation of a 6-week gestational limit on abortion in Texas in 2021.^8,9 Moreover, requests for self-managed medication abortion through the online telemedicine service Aid Access showed sharp increases immediately after Dobbs, especially in states with bans.¹⁰ Yet beyond increased demand, little else is known about the number of provisions of medication for self-managed abortion in the wake of Dobbs.

This study directly addresses this gap in knowledge. By compiling data from as many known sources as possible, we counted how many provisions of medications for self-managed abortion took place during the 6 months immediately after Dobbs. We then adjusted this raw count in 2 ways: by subtracting what might reasonably have been expected based on the pre-Dobbs status quo and by accounting for the possibility that not all provided medications are used by their recipients.

We sought to obtain data from as many entities as possible that provided abortion medications outside the formal US health care setting between March 1 and December 31, 2022. To do so, we conducted a landscape review that encompassed internet searches; consultation with activists, advocacy organizations, and journalists; and information provided by Plan C, an organization that provides up-to-date information on how people in the US can access self-managed medication abortion. We identified 3 types of sources: (1) telemedicine organizations providing abortion medications with the involvement of non鈥揢S-based clinicians; (2) community networks providing abortion medications through volunteers; and (3) online vendors selling abortion medications. Each type of source uses a different model to provide medications for self-managed abortion. Telemedicine organizations typically involve a physician located outside the US who prescribes mifepristone and misoprostol. The recipient is asked to report relevant clinical history and donate approximately $100 to the service, with reduced donation amounts based on financial need. Medications are then mailed to the US from an overseas pharmacy. Instructions and support are provided by an online help desk.⁷ Community networks typically involve medications sourced from pharmacies outside of the US, which may be mailed or hand delivered to recipients by a network of volunteers who also provide peer-to-peer support and information. Similar information and support may also be provided through hotlines. These services often provide medications at no cost to recipients.¹¹ Online vendors typically involve a website interface through which people can purchase abortion medications for between $39 and $470. Medications are then mailed to recipients from either within or outside the US, typically without integrated provision of information or support.¹²

We asked all identified sources to share data on monthly provision volume (defined as the number of provisions of abortion medications) over 2 periods: March 1 to April 30, 2022 (baseline) and July 1 to December 31, 2022 (post-Dobbs). For sources operating as umbrella organizations comprising multiple subsources (ie, online vendors operating multiple websites, community networks with multiple interconnected national and/or international branches), we ensured that data were aggregated across subsources in a coordinated fashion and that no duplicate reporting occurred. We asked sources that involved both US- and non鈥揢S-based clinicians, such as Aid Access, to exclude provisions by US-based physicians. During the study period, these provisions were limited to states with laws permitting telemedicine for medication abortion. We also asked sources that offer advance provision, ie, provisions of abortion medications for possible later use by people who are not pregnant, to exclude these provisions from their reported volumes.¹³ We excluded May and June 2022 from the study period for 2 reasons: (1) the extreme volatility of demand between the leak of the Dobbs decision and its formal announcement¹⁰ and (2) the fact that these months are not included in published estimates of abortions provided within the formal health care setting.⁴ To respect requests for confidentiality, we aggregated counts at the level of source type rather than individual source.

For each source type, we counted the total number of provisions of abortion medications in each of the 6 months after the Dobbs decision. We then compared these observed monthly provisions with the 2 months of monthly provision counts in the pre-Dobbs data, which allows estimation of what might have been observed in a counterfactual scenario in which Dobbs had not occurred. This comparison is subject to uncertainty because the pre-Dobbs counts can provide only an estimate of what might have occurred in the absence of Dobbs. To account for this uncertainty, we made 2 key assumptions. First, we modeled the pre-Dobbs monthly counts using a negative-binomial distribution with a constant monthly mean specific to each source type. We fit this model in a bayesian framework using noninformative prior distributions, and we verified the fit of the model using posterior predictive checks. Second, we assumed that if Dobbs had not occurred, the post-Dobbs provision counts would have followed the same negative-binomial distribution that we used to model to the pre-Dobbs counts. Based on these 2 assumptions, we used our fitted model to repeatedly simulate counterfactual post-Dobbs counts from the posterior predictive distribution, and we subtracted these simulated counterfactual counts from the actual post-Dobbs counts. This gave us a posterior distribution over excess provisions: the difference between what was actually observed in July through December 2022 and what might have been observed over that same period had Dobbs not occurred. Full details of the model are in the eAppendix in Supplement 1.

To estimate the change in actual self-managed medication abortions taking place after Dobbs, we multiplied excess provisions by an assumed usage rate, which represents the fraction of medications provided that are actually used by their recipients. We estimated the usage rates separately by source type using data from the sources in our dataset cross-referenced with published estimates. The telemedicine organizations in our dataset reported an average usage rate of 88% based on follow-up with their recipients. We cross-referenced this number with 5 independent published studies of self-managed medication abortion provided by telemedicine, which yielded very consistent estimates: 86%, 87% (2 studies), 88%, and 90%,^14-18 for an average of 88%. Similarly, we used the average of 4 recent articles on community networks, which had very consistent estimates of the usage rate, 97% and 99% (3 studies),^19-22 yielding an average of 98.5%. Although community networks in our dataset were unable to provide data on usage, the estimate of 99% is consistent with their best estimates. We could find no prior work on the usage rate of abortion medications sold by online vendors, and these sources were not able to supply data on usage. However, we judged that they are more similar to telemedicine services than they are to community network models. Therefore, we used the lowest available published usage rate from the 4 studies on telemedicine, yielding an assumed usage rate of 86% for online vendors.

Although we could not obtain provision volumes from a small number of online vendors identified in the landscape review, we did obtain Google Analytics data on outbound clicks to these vendors鈥� websites originating from Plan C鈥檚 website.²³ Crucially, we also had data for a set of online vendors that were 鈥渇ully observed鈥� in the sense that both monthly outbound clicks and monthly provision volumes were available. We therefore fit a hierarchical bayesian regression model to these vendors鈥� data, allowing formation of posterior distributions of missing values for vendors with observed clicks (x) but unobserved provision volumes (y). Results are reported herein both with and without these imputed values. Details of the imputation model are in the eAppendix in Supplement 1.

Sources did not share any personal or potentially identifying data with the research team. The University of Texas Institutional Review Board approved the study. All data analysis was conducted using R version 4.3.1.²⁴ We followed the STROBE reporting guidelines for observational studies.

We identified 15 unique sources, encompassing 28 subsources, that provided medications for self-managed abortion and that operated between March 1 and December 31, 2022. To our knowledge, this sample includes most, if not all, active sources during the study period. Among the 15 unique sources, 11 shared data on monthly provision volumes, including all the telemedicine organizations and community networks included in the sample. Four of 11 unique online vendors did not share data; for these sources, we imputed counts using the bayesian regression model for provision volume (y), given outbounds clicks (x) to each vendor鈥檚 website.

The Figure shows reported monthly provisions, separately by source type, during the pre-Dobbs and post-Dobbs periods. Excluding imputed missing values, there was an average of 5931 counted provisions per month totaled across all source types reported during the 6-month post-Dobbs period; the pre-Dobbs average was 1407. During the post-Dobbs period, community networks accounted for 3054 provisions per month (51.4%); telemedicine organizations operating outside the formal health care setting, 2204 provisions per month (37.2%); and online vendors, 674 provisions per month (11.4%).

The Table shows the estimates for excess post-Dobbs provision volume derived from the bayesian model, separately by source type. Totals are presented both with and without the imputed data. Including imputed counts, we estimated that the 6-month post-Dobbs period saw an increase in provisions of 27鈥�838 (95% credible interval [CrI], 26鈥�374-29鈥�175) compared with what our model would expect based on the pre-Dobbs data. Excluding imputed counts changes the estimate of excess provisions only slightly (27鈥�145; 95% CrI, 25鈥�747-28鈥�246).

The Table also shows the estimated change in self-managed medication abortions after adjusting excess provisions to account for the estimated usage rates for each source type. Including imputed counts, we estimated a 6-month post-Dobbs increase in self-managed medication abortions of 26鈥�055 (95% CrI, 24鈥�739-27鈥�245) vs what would have been expected from the pre-Dobbs baseline.

This study provides estimates of the change in provision of medications for self-managed abortion, as well as actual use of these medications, in response to the Dobbs decision and the subsequent implementation of state-level abortion bans. We estimated that 27鈥�838 additional provisions of abortion medications and 26鈥�055 additional self-managed medication abortions took place in the 6 months following Dobbs compared with what might have reasonably been expected under the status quo. The results are robust to the imputation of missing values.

Counting self-managed medication abortions poses many challenges. By definition, these abortions take place outside the formal health care setting and therefore create no formal administrative records. These events are also difficult to count in surveys: underreporting of abortion by survey respondents is a well-established issue,²⁵ one likely to be exacerbated by the private nature and potential legal risks of self-managed abortion. The method used in this study, whereby data were obtained directly from sources that provide abortion medications outside the formal health care setting, was specifically undertaken to circumvent these challenges.

It is possible that our estimate may undercount the true number of provisions of medications for self-managed abortion during the post-Dobbs period. Indeed, we expected this to be the case: while we have tried to be as comprehensive as possible, not all sources of abortion medication outside the formal health care setting could be included. Notably, the count does not include provisions by physical pharmacies in Mexico, where people can obtain misoprostol after crossing the US border, as well as provisions at community settings such as flea markets in border towns²⁶ and provisions by community home abortion providers.²⁷ We could find no published estimates of how common these practices are, but in the context of the entire US, most of which is far from the US-Mexico border, it seems unlikely that these pathways account for a large fraction of all provisions. Although we expected these facts to bias our estimate downward, it seems unlikely that we have drastically undercounted. Alternatively, it is possible that our estimate may overcount the true number of provisions of medications for self-managed abortion during the post-Dobbs period. Our data reflect the provision, but not necessarily the use, of abortion medications for self-management. Although the analysis corrected for this fact by using the best available published estimates of the usage rate, if these estimates of the usage rate are biased upward, then so too is the estimate of how many abortions took place. Studies of both self-managed and clinical abortion often include people to whom medications were provided but whose outcomes are unknown, and we can only estimate usage based on available data from people who provide follow-up information about medication receipt and use. However, additional contact with people initially considered lost to follow-up due to nonresponse yields similar usage estimates to those used in our analysis.¹⁵

There are several limitations to this study. First, we were unable to cross-reference provisions across different sources, leaving open the possibility that distinct shipments from 2 or more sources were actually for the same person. However, while people may order abortion medications from multiple sources, this issue is at least partially mitigated by adjusting for a usage rate of less than 100%. Indeed, in the literature that characterizes usage rates, one of the main reasons people offer for not using provided medications is that they obtained an abortion from another source.^16,17,19 It also seems unlikely that multiple ordering would be common given that most source types require payment or a donation, and cost can represent a significant barrier for people to access even a single source of self-managed medication abortion.²⁸ Second, while sources that offer advance provision explicitly excluded these shipments from their reported volumes,¹³ sources that do not ask about clinical history, namely online vendors, may have been unable to distinguish these provisions from provisions intended for immediate use.

Third, we were unable to verify data from all sources, since some do not keep formal records. Fourth, we were able to obtain only 2 months of pre-Dobbs data from most sources; however, the bayesian model naturally accounts for the uncertainty arising from the limited size of the dataset, which is reflected in the posterior summaries in the Table.

Our findings suggest that even though fewer people accessed abortion care within the formal health care setting in the 6 months after Dobbs, a substantial number were able to access abortion medications outside the formal health care setting, despite state-level bans and restrictions. While self-managed medication abortion has long been considered a marginal practice in US, our findings suggest that this approach has become mainstream. This phenomenon is consistent with trends observed globally: in countries where access to in-clinic abortion is banned or heavily restricted, people frequently self-manage instead.²⁹ Indeed, the types of community networks now serving the US originated from the model of accompaniment networks in Latin America,³⁰ and the online telemedicine services operating in the US are modeled on international organizations like Women on Web.³¹ The public health implications of this shift are complex. Self-managed medication abortion provided using telemedicine and community network models is known to offer high rates of effectiveness and very low rates of serious adverse outcomes.^5,15-18,32 Many people find it not merely acceptable but actively preferable to clinical care.³³ However, in the face of abortion bans, those who self-manage may face legal risks³⁴ and/or experience difficulty accessing postabortion care within the formal health care setting, should they want or need it.³⁵

Self-managed medication abortion is not a familiar or feasible option for all. Accordingly, it is likely that a substantial number of people continued their pregnancies.³⁶ Indeed, data on birth counts for the first half of 2023 suggest an increase of approximately 2.3% in states with total abortion bans in place relative to states without such restrictions.³⁷ This increase in births in ban states coincides with an increase in abortions taking place within the formal health care setting in states without bans or severe restrictions. Although such abortions declined nationally during our study period, during the immediate aftermath of Dobbs, data gathered over subsequent months have indicated a rebound in formal health care setting abortions during 2023.^38,39 The fact that this increase is concentrated in states without bans suggests that it is likely the result of increased delivery of medication abortion via telehealth, which has expanded access by reducing financial and logistical burdens for many. While declines in formal health care setting abortions have persisted in states with bans, recent developments such as provision of medication abortion to people residing in these states under shield laws may reverse this trend. Further research is needed to determine the relative contributions of various service delivery models in restrictive states. In the meantime, self-managed medication abortion will likely remain a significant contributor to abortion access in the post-Roe US. Clinicians will need to be prepared to encounter patients who may be considering self-managed medication abortion or who need or want to connect with the formal health care setting for postabortion care.

Provision of medications for self-managed abortions increased in the 6 months following the Dobbs decision. Results suggest that a substantial number of abortion seekers accessed services despite the implementation of state-level bans and restrictions.

Corresponding Author: Abigail R. A. Aiken, PhD, LBJ School of Public Affairs, University of Texas at Austin, PO Box Y, Austin, TX 78712 (araa2@utexas.edu).

Accepted for Publication: March 6, 2024.

Published Online: March 25, 2024. doi:10.1001/jama.2024.4266

Author Contributions: Dr Aiken had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Aiken, Scott.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Aiken.

Critical review of the manuscript for important intellectual content: All authors.

Statistical analysis: Scott.

Obtained funding: Aiken.

Conflict of Interest Disclosures: Dr Aiken reported receipt of honoraria from RAD: Resources for Abortion Delivery for presentation of research on self-managed abortion and from Mathematica Inc for consultation on a proposed policy evaluation project relating to abortion. Dr Gomperts reported being founder and director of Aid Access. No other disclosures were reported.

Funding/Support: Dr Aiken receives grant support from the Society of Family Planning (grant SFPRF12-MA1), the Kopcho Reproductive Freedom Fund, and the William and Flora Hewlett Foundation (grant 2023-02900-GRA) and infrastructure support from the National Institutes of Health (grant P2CHD042849).

Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; or decision to submit the manuscript for publication.

Data Sharing Statement: See Supplement 2.