key: cord-0772879-b42xa16m
authors: Wu, Chenhui; Cheng, Jianquan; Zou, Jun; Duan, Lian; Campbell, Janis E.
title: Health-related quality of life of hospitalized COVID-19 survivors: an initial exploration in Nanning City, China
date: 2021-02-12
journal: Soc Sci Med
DOI: 10.1016/j.socscimed.2021.113748
sha: b9df9ab0b9bf713e26766384bcef99611d6a51a2
doc_id: 772879
cord_uid: b42xa16m

Understanding the health-related quality of life (HrQoL) of hospitalized COVID-19 survivors is an emerging global challenge arising from the current pandemic. A qualitative study of the experiences of sixteen hospitalized COVID-19 survivors from Nanning City, China, was conducted using semi-structured telephone interviews in May 2020. These first-hand accounts were critically and empirically analysed to identify emerging health and social issues, and provide potential solutions to improve survivors’ quality of life. This in-depth, qualitative study of HrQoL for hospitalized COVID-19 survivors provides the first empirical evidence and conceptual framework with eight dimensions (physical symptoms, anxiety, trauma, economic loss, placebased identity, self-stigma, health self-interventions, and changing lifestyle) for understanding their physiological, psychological, socio-economic and health behavioral aspects of the daily lives. We argue that local and global governments should provide integrated healthcare, social and digital infrastructure to support this vulnerable group. More comparative and multi-disciplinary studies in this area as needed to generate academic standards of assessing health-related quality of life and produce good practice guidelines for promoting urban resilience in response to public health disasters.

One study found that after clinic closures caused by admitting privilege and ambulatory surgical center laws in Texas, the number of abortions in the state decreased by 13%, and the number of second trimester abortions increased (Grossman et al., 2014) .

Taken together, restrictive state-level policies create multiple social, logistical and financial barriers to abortion, and in some cases make it impossible for people to obtain an abortion. Various abortion restrictions, such as bans on insurance coverage, gestational limits, and TRAP laws, increase both the number of people who need to travel to access abortion, as well as the distance they must travel (Barr-Walker et al., 2019) . Having to travel long distances for an abortion can lead to delays in abortion care, higher costs, and increased emotional burdens (Barr-Walker et al., 2019; Jerman et al., 2017) . Many of these burdens are exacerbated for those who are young, rural, low income, or over 12 weeks gestation in pregnancy (Barr-Walker et al., 2019, p.; Jerman et al., 2017; Margo et al., 2016; White et al., 2016) . Other logistical barriers, such as insurance difficulties or not knowing where to go or how to get to a provider, as well as financial barriers, such as the need to raise money for travel and procedure costs, are common reasons for delay in abortion seeking (Upadhyay et al., 2014) .

The financial costs of abortion, in particular, create the greatest obstacles to abortion care. Many private health insurances do not cover abortion and many abortion patients report difficulty affording their abortion, sometimes delaying or not paying other bills in order to pay for abortion (Jones et al., 2013) . These challenges are often insurmountable-for example, among a group of women denied an abortion because of gestational age limits who considered going to another clinic, 85% reported procedure and travel costs as the primary reason for not obtaining an abortion elsewhere (Upadhyay et al., 2014) .

Lack of Medicaid coverage of abortion exacerbates financial barriers for low-income people. The Hyde Amendment bans federal Medicaid coverage of abortion except in cases of life endangerment, rape, or incest. However, 16 states have opted to use their own funds to J o u r n a l P r e -p r o o f cover abortion for those eligible for Medicaid (Salganicoff et al., 2020) . For low-income individuals in the remaining 34 states plus Washington, DC, paying for abortion can be a catastrophic health expenditure (Zuniga et al., In press) . Individuals often must resort to borrowing money or relying on local donation-based funds that provide financial and logistical assistance to people seeking abortion (Ely et al., 2017) . Not everyone is able to raise the money, though. Previous studies indicate that lack of state Medicaid coverage of abortion makes it impossible for some people to obtain wanted abortions (Cook et al., 1999; Roberts et al., 2019) .

A major challenge in understanding barriers to obtaining abortion is that many of those who are most affected may never reach an abortion provider, making the full impact of restrictive policies on people's experiences difficult to measure. Most studies about the impact of state-level restrictions and other barriers to abortion care have been conducted among abortion clinic patients, which limits our knowledge to the experience of those able to reach abortion services. Less is known about individuals who want an abortion but never actually present for care. Those who do not reach a clinic may differ in key ways from those who reach a clinic and may face additional structural barriers to abortion access.

To address this concern, a few studies have used innovative approaches to explore the experiences of women who considered abortion but may not have presented at an abortion clinic O'Donnell et al., 2018; Roberts et al., 2020) . Fuentes et al. (2016) interviewed people who were unable to access an abortion clinic (could not make appointments or had existing appointments canceled) after the passage of House Bill 2 closed many Texas abortion clinics . Due to clinic closures, women experienced barriers that often delayed abortion care and even prevented some people from obtaining an abortion at all.

In their qualitative research in rural Appalachia, O'Donnell et al. (2018) interviewed women recruited in multiple settings, including at centers of commerce, to overcome traditional J o u r n a l P r e -p r o o f sampling limitations and include those who may have considered, but not sought, an abortion (O'Donnell et al., 2018) . Those who experienced unwanted pregnancies did not always seek abortion, and described a complex decision-making process shaped by the options that were feasible. Roberts et al. (2020) conducted a mixed methods study among women entering prenatal care in Louisiana and Maryland and found that 30% had considered abortion for their current pregnancy and that 6% of those in Louisiana and 1% in Maryland reported a policyrelated barrier to obtaining an abortion, primarily lack of Medicaid coverage (Roberts et al., 2020 (Roberts et al., , 2019 . In this study, higher levels of economic insecurity were associated with experiencing a policy-related barrier to obtaining an abortion (Roberts et al., 2020) .

These studies provide key insights into the experiences of those who consider, but do not obtain, an abortion. However, no studies have systematically focused on surveying women while they were in the process of seeking an abortion before reaching an abortion clinic to investigate the extent to which abortion laws affect their ability to obtain a wanted abortion.

Internet recruitment is a new strategic way to reach individuals who are considering abortion.

For this study, we utilized an innovative methodology that recruited individuals from throughout the U.S. searching the Internet for abortion care to better understand how the abortion policy climate and state Medicaid coverage of abortion were associated with ability to obtain an abortion. Using this recruitment method allowed us to reach individuals who may face barriers to accessing abortion services and never reach a clinic, thus reaching a broader population in abortion research. This study aimed to compare pregnancy outcomes by state policy climate and by state Medicaid coverage of abortion among a population considering abortion. We also aimed to describe financial and logistical barriers to abortion that people face in obtaining abortion care.

The Google Ads Abortion Access Study is a national longitudinal cohort study that recruited people searching online for abortion care between August 2017 and May 2018. The study was designed to understand how state-level abortion policies impact individuals' ability to obtain a wanted abortion and to understand barriers and facilitators to a wanted abortion. The study recruited individuals searching Google for an abortion provider. Advertisements for the study were displayed at the top of search results and candidly advertised the opportunity to participate in a "University Study" on abortion with compensation of $50. We used a stratified sampling strategy to ensure broad national representation, with an initial goal of recruiting 20 participants from every state plus Washington, DC to capture experiences of those in less populated states. To do this, we monitored the enrollment and follow-up of participants for each state daily and when we met our goal, we stopped running ads in that state. This ensured that the entire sample was not made up of only participants from the most populated states.

However, in 8 states we recruited fewer than 20 (range: 8-18).

Users who clicked on the advertisement were directed to a landing web page explaining the study and then to a screening form. People were eligible if they reported being pregnant and currently considering abortion. They then completed an online baseline survey and 4 weeks later, were invited by email or text message to complete a follow-up survey. Participants who completed the follow-up survey were remunerated with a $50 gift card. Further detail on the methods of this study are reported elsewhere (citation removed to preserve anonymity).

Parental consent for minors was waived because 1) all study participants were pregnant and able to consent to their own medical care and thus also able to consent to participate in the research study, and 2) the research protocol was designed for a population for which parental or guardian permission was not a reasonable requirement in order to protect the participants.

J o u r n a l P r e -p r o o f

The primary outcome for this study was pregnancy outcome, a 3-level indicator of pregnancy status as reported at the 4-week follow-up time point: 1) had an abortion, 2) pregnant but still seeking an abortion, or 3) pregnant and planning to continue pregnancy. We chose to follow up at 4 weeks because we hypothesized it would give most participants enough time to locate and visit an abortion provider for those who wanted to, yet still be soon enough for the participant to remember the process they went through, reducing recall bias.

The 2 primary independent variables of interest were the state policy climate and state Medicaid coverage of abortion. We used NARAL Pro-Choice America's state access ratings as a proxy for the state policy environment for abortion (NARAL Pro-Choice America, 2018).

NARAL rates each state annually based on the status of reproductive healthcare access, including policies related to contraception and abortion access using 5 categories, which we collapsed into 3 classifications: 1) protected access (strongly protected access and protected access), 2) some access, and 3) restricted access (severely restricted and restricted access) ( Figure 1 ). Each participant's state access rating was determined based on their reported state of residence and the year they enrolled in the study. We determined which states cover abortion using state Medicaid funding at the time of the study ( Figure 2 ).

Other key variables were the reported barriers experienced during the process of obtaining or trying to obtain an abortion. We defined barriers as things that made getting an abortion harder or prevented participants from getting an abortion. To assess participants' experience of barriers, the follow-up survey asked, "Did any of the following things make it hard to get an abortion?" followed by a list of potential barriers, among which they could select all that applied. While multiple personal, financial, and logistical barriers were listed, in this analysis we focused only on financial and logistical barriers, such as needing to obtain financial resources to J o u r n a l P r e -p r o o f get an abortion or having to arrange transportation to get to an abortion clinic, because these factors are most related to state abortion policies.

The follow-up survey also asked individuals who obtained an abortion about facilitators:

"Did any of the following things make it easier for you to get an abortion?", followed by a list of potential facilitators, among which they could select all that applied.

For participants who were still pregnant at follow-up, the survey asked about their efforts to obtain an abortion, including whether they tried to get an abortion (allowing respondents to define "tried" for themselves), called an abortion clinic, made an appointment at an abortion clinic, or visited an abortion clinic. Finally, participants were asked, "At this point, do you wish you had had an abortion?" Self-reported sociodemographic characteristics measured at baseline included the participants' age, race/ethnicity, highest education level completed, and employment status.

Race/ethnicity was included as a proxy for the differences in social, environmental, and structural factors individuals of different races/ethnicities may face, such as racism. We asked participants how often in the last 12 months they had enough money to meet basic living needs, which we reverse coded and dichotomized; those who reported "some of the time, rarely or never" were coded as having "difficulty meeting basic needs." We also asked about participants' relationship to the other person involved in the pregnancy (current partner, former partner, or other), insurance type, and religiosity. Pregnancy gestation at baseline was calculated based on self-reported last menstrual period and categorized into five levels, including a missing category.

Finally, we included a validated measure of decision conflict about the abortion assessed at baseline with the question format of the Decisional Conflict Scale (DCS) (O'Connor, 1995) . The DCS is considered the gold standard for assessing conflict in the context of health care decisions and has been previously used in abortion research . Scores range from 0 to 100 and previous studies have found that scores <25 have been associated with J o u r n a l P r e -p r o o f success in implementing decisions, while scores >37.5 have been associated with delayed decision making or uncertainty about implementing a decision (O'Connor, 1993) . Thus, we categorized the scores into low conflict (<25), medium conflict (25-37.5), and high conflict (>37.5).

First we described the sociodemographic characteristics of the sample and compared these characteristics by the state abortion access rating using chi-square tests. Second, we described the types of attempts made to get an abortion among those who were still pregnant at follow-up, comparing the proportions of those who were still seeking an abortion to those who were planning to continue the pregnancy.

Third, we used two separate multilevel multinomial logistic regression models to estimate the 3-level pregnancy outcome dependent variable, with the group that obtained an abortion as the reference category. The first model included state abortion access rating and the second included state Medicaid coverage of abortion as the main independent variables. For both we derived our estimates by using a mixed-effects model with multinomial distribution and logit link function. Multilevel models were used to account for lack of independence among participants from the same state while assessing the fixed effects of sociodemographic and pregnancy-related variables, with the participant's state of residence entered as a random effect. Results of the mixed-effects models are expressed with adjusted odds ratios (aORs) and 95% confidence intervals (CIs).

Fourth, we described reported barriers to abortion for those who were still pregnant and still seeking an abortion and for those who were planning to continue their pregnancies, assessing potential differences by state access rating using chi square tests. We then described barriers and facilitators to abortion among those who obtained an abortion, by state access rating, again assessing differences by state access rating using chi square tests. Finally, in a J o u r n a l P r e -p r o o f post-hoc test, we examined whether living in states with Medicaid coverage of abortion was associated with reporting financial-related barriers to abortion. All statistical tests in this study were two-tailed with a level of significance set at 0.05. Analyses were conducted using Stata, version 15 (Stata Corporation, College Station, TX, USA).

In addition to the quantitative survey measures described above, all survey participants

were asked several open-ended questions. Those who had obtained an abortion were asked, "What were the main thing(s) that made it harder to get an abortion?" Respondents who were still pregnant at the time of the follow-up survey were asked, "What were your main reason (s) for not getting an abortion?" To code these open text responses, two study researchers looked at the responses for trends and grouped responses into categories. Initial codes were discussed with the research team and revised as needed. After refining and finalizing the codes, a coauthor (AB) and a research assistant independently coded the responses for each of the responses for these 2 questions. Any discrepancies between coders were resolved through discussion and consensus. We then purposefully selected quotes from the codes labeled "costs/insurance/logistics" and "gestational age," which were the most commonly reported policy-related barriers.

A total of 1,982 people were eligible, consented to participate, and started the baseline survey, and 1,485 (75%) of the cohort completed the baseline survey and provided contact information for follow-up. We removed 3 participants who reported living outside the US and 21 who were found to have made multiple attempts to take the survey or were referred to the survey from an external site. A total of 1,005 completed the follow-up survey for a 69% follow-J o u r n a l P r e -p r o o f up rate. Older age, white race, higher education, not having difficulty meeting basic needs, being not religious/spiritual and having no previous births were associated with higher follow-up rates (citation removed to preserve anonymity). Follow-up rates did not differ by state abortion access rating or state Medicaid coverage of abortion.

Among those who provided follow-up data, we excluded 86 reporting a stillbirth or miscarriage, 25 reporting never being pregnant, 14 reporting having had a live birth since the baseline survey, and 6 whose outcomes were unknown, leaving a final analytic sample of 874.

Of the 874 participants, 237 (27%) lived in protected access states, 109 (13%) lived in states with some access, and 528 (60%) lived in states categorized as having restricted access (Table 1 ). There were significant differences in race/ethnicity across the state groups (p=0.005).

A higher proportion of the sample identified as Black in restricted access states (28%) than protected access states (17%).

Participants living in protected access states were more likely to have Medicaid, Medicare, or state exchange insurance coverage (58%) than those living in restricted access states (49%), while participants living in protected access states were less likely to be uninsured or have unknown insurance (19%) than those living in restricted access states (27%) (p=0.001).

Among all participants, 71% lived in states that did not offer Medicaid coverage of abortion. The only characteristic that differed significantly between states that offered coverage of abortion and those that did not was type of health insurance. Participants living in states that provide Medicaid coverage of abortion were significantly less likely to be uninsured or have unknown insurance (16%) than those living in states that do not provide Medicaid coverage (26%) (p=0.003).

At follow-up, 48% of the sample reported that they had had an abortion, 32% were still seeking an abortion, and 20% were planning to continue their pregnancies. Table 2 shows the efforts made to obtain abortion by those still seeking an abortion at follow-up compared to those continuing their pregnancies. Overall, 57% of the still seeking group reported that they tried to get an abortion, compared to 35% in the continuing pregnancy group. Those who were still seeking an abortion had significantly higher rates of all steps taken, but a sizable proportion of those planning to continue their pregnancy also made attempts to obtain an abortion, including 63% who reported calling an abortion clinic. Almost two-thirds (60%) of those still seeking an abortion and 14% of those continuing their pregnancies reported at follow-up that they wish they had had an abortion (p<0.001). Additionally, 44% of those still seeking an abortion reported having an appointment scheduled at the time of the survey.

The multivariable analysis found that at follow-up, participants living in restrictive access states had higher odds of being pregnant and planning to continue the pregnancy (versus having an abortion) compared to participants in protective access states (aOR=1.70, 95% CI=1.08, 2.70) ( Table 3) . Individuals living in restricted access states trended towards higher odds of being pregnant and still seeking an abortion (versus having had an abortion) compared to those in protected access states, although this association did not reach statistical significance (aOR=1.44, 95% CI=0.99, 2.09).

Relative to white participants, Black participants had higher odds of being pregnant at follow-up and still seeking an abortion (aOR=1.51, 95% CI=1.01, 2.24) and of planning to continue the pregnancy (aOR=1.78, 95% CI=1.13, 2.80). Latinx participants also had higher odds of being pregnant and still seeking an abortion at follow-up (aOR=1.69, 95% CI=1.02, 2.80).

J o u r n a l P r e -p r o o f Participants greater than 14 weeks pregnant at baseline had greater odds of being pregnant and still seeking abortion at follow-up (versus having an abortion) than those who were ≤10 weeks pregnant at baseline (aOR=3.27, 95% CI=1.47,7.27). Reported gestation of pregnancy at baseline was also associated with being pregnant and planning to continue the pregnancy (versus having an abortion) at follow-up: 10 to 14 weeks (aOR=2.53, 95% CI=1.48,4.36), 14 to 20 weeks (aOR=4.90, 95% CI=2.09,11.52), and 20 weeks and greater (aOR=7.13, 95% CI=1.77,28.77). Finally, greater decision conflict at baseline was associated with being pregnant at follow-up: those with medium conflict (aOR=1.51, 95% CI=1.00, 2.28) and high conflict (aOR=2.53, 95% CI=1.71, 3.73) had higher odds of still seeking an abortion at follow-up. Similarly, those with medium (aOR=2.01, 95% CI=1.25,3.24) and high conflict (aOR=3.13, 95% CI=1.99,4.93) had higher odds of planning to continue the pregnancy at followup. Baseline characteristics not associated with pregnancy outcome were age, education, employment status, difficulty meeting basic needs, type of health insurance, and religiosity.

The second multivariable model demonstrated that individuals in states that do not provide Medicaid coverage of abortion had significantly higher odds of being pregnant and still seeking an abortion at follow-up versus having had an abortion (aOR=1.80, 95% CI=1.24,2.60).

State Medicaid coverage of abortion was not associated with being pregnant and planning to continue the pregnancy (versus having had an abortion) at follow-up. All associations of covariates found in the previous model remained similar in this model (Table 4) .

Overall, policy-related barriers to abortion were common and reported by all participants, including those who obtained an abortion. The most common barriers were having to gather J o u r n a l P r e -p r o o f money for travel expenses or to pay for the abortion (67%), having to take time off work and school (47%), having to figure out if insurance would cover the abortion (44%), and how far along in the pregnancy the participant was (41%) (not shown).

In examining barriers among individuals by pregnancy outcome, for those who were pregnant but still seeking an abortion at follow-up, having to gather money for travel expenses or to pay for the abortion was by far the most commonly reported barrier (80%) ( Table 5) .

Individuals in restricted access states were more likely to report money as a barrier (87%) than those in protected access states (65%) (p<0.001). Participants living in restricted access states were more likely than those living in other states to report having to travel a long distance as barriers to abortion care (p=0.012) and having to arrange for child or other dependent care (p=0.024). There were no other statistically significant differences in barriers by state access rating among those still seeking abortion at follow-up.

Among individuals planning to continue the pregnancy, the most commonly reported barrier was also having to gather money for abortion-related expenses (52%) ( Table 6) .

However, there were no significant differences in other reported barriers by state access rating among those planning to continue the pregnancy.

Individuals who had obtained an abortion at follow-up also reported barriers to care.

Similar to the other groups, having to gather money to pay for travel expenses or for the abortion was the most commonly reported barrier, cited by 64% of participants who had an abortion at follow-up, and was more frequently reported by those in restricted access states (71%) than protected access states (53%, p=0.003) ( Table 7) . Participants living in restricted access states were more likely than those living in other states to report having to make multiple trips to the clinic (p=0.002) and protestors at the clinic (p=0.002) as barriers. Those in states with some access were more likely to report having to get time off work/school than those in J o u r n a l P r e -p r o o f protected access states (p=0.044) and more likely to report having to figure out if their insurance would cover the abortion than those in restricted access states (p=0.021).

Individuals who obtained an abortion also reported facilitators to care. Being able to easily find information online about where to go to get an abortion was the most commonly reported facilitator to obtaining an abortion, cited by 86% of participants who had an abortion by follow-up, with no differences by state access rating (Table 7) . Participants in protected access states were more likely than those in other states to report that insurance coverage of their abortion and that the abortion clinic being close to where they lived acted as facilitators.

Participants in restricted access states were significantly more likely to report someone giving or lending them money for the abortion (p<0.001) and the clinic helping them find the money to pay for the abortion as facilitators.

To elucidate the finding that having to gather money for travel expenses or to pay for the abortion was the most commonly reported barrier among all groups, we examined the prevalence of this barrier by state Medicaid coverage of abortion. Over 71% of participants lived in states where Medicaid did not cover abortion care (not shown). Individuals living in these states without Medicaid coverage of abortion were more likely than those in Medicaid coverage states to report having to gather money to pay for travel expenses or the abortion as a barrier to abortion care (78% vs. 55%, p<0.001) (Figure 3 ).

Reviewing open-ended responses to the question about barriers to abortion access among those who were still pregnant at follow-up echoed many of the themes endorsed in Tables 5 and 6. Participants who were pregnant and still seeking abortion and those continuing their pregnancies described a variety of reasons for not having had an abortion. Examples of financial and logistical barriers include:

• "The price is so high in Las Vegas, NV I'm trying to collect the funds before I'm 13 weeks. The state of NV also does not have financial aid for abortion available like some other states .... Price is the only obstacle at this point in time, and the longer you wait the more expensive it becomes which is devastating." -Age 29, multiracial, ≤10 weeks pregnant, still seeking abortion, protected access state, no Medicaid coverage state

• "I just found out my insurance doesn't cover it, so I will have to come up with the money." -Age 35, Black, ≤10 weeks pregnant, still seeking abortion, restricted access state, no Medicaid coverage state

• "I haven't yet gotten an abortion because with our four young children, all under age of four, we can't afford the cost and is so much $ [money] for the procedure everywhere and they don't accept insurance, even with the clinic helping me with some of the cost because I qualify for it, it is so hard to come up with this money. Last week we actually were able to get up most of the $ [money] and they were going to do it so I had an appointment and everything, but the Category 5 Hurricane … was coming straight for us and I didn't feel safe with the babies in our home so we had to leave and spend the $ [money] on the hotel for few days and everything else. 

In this study we found that living in a U.S. state with a restrictive abortion policy climate was associated with continuing pregnancy after considering abortion. This study is consistent with previous study findings that state-level abortion restrictions create numerous financial and logistical barriers to abortion (Cohen and Joffe, 2020; Fuentes et al., 2016; Jerman et al., 2017; Roberts et al., 2020 Roberts et al., , 2016 White et al., 2016) . This study builds on the existing evidence using a large national sample of people considering or seeking abortion and shows that abortion restrictions are associated with some ultimately not having abortions at all.

Additionally, living in a state that does not offer Medicaid coverage of abortion was associated with prolonged abortion seeking, as evidenced by higher odds of still seeking abortion at 4 weeks follow-up among individuals living in these states. This is consistent with previous research demonstrating that lack of Medicaid coverage creates barriers to and delays care (Cook et al., 1999; Roberts et al., 2019) . While almost half of those still seeking an abortion had an appointment scheduled, 4 weeks is still long to wait for an abortion and an indicator of poor quality of care (National Academies of Sciences, Engineering, and Medicine, 2018).

Furthermore, many of those in this group may end up not having an abortion at all.

People seeking abortions are already disproportionately economically disadvantaged compared to the general population (Jerman et al., 2016) and previous research has shown that those who are economically disadvantaged are more likely to report that abortion policies have been barriers to abortion care (Roberts et al., 2020) . In addition, the longer people seeking abortion are delayed, the more expensive the abortion, leading to a perpetual cycle of fundraising until the clinic or state gestational limit is reached and it is too late for an abortion (Upadhyay et al., 2014) . These findings indicate that restoring federal Medicaid coverage of abortion by repealing the Hyde Amendment would remove an often insurmountable barrier to J o u r n a l P r e -p r o o f abortion care for low-income people regardless of the state they live in. Research indicates that Medicaid coverage bans contribute to approximately 1 in 4 low-income pregnant people giving birth instead of having an abortion when Medicaid does not cover abortion (Henshaw et al., 2009; Roberts et al., 2019) .

There is likely a complex relationship between abortion accessibility and decisionmaking. While a variety of issues contribute to the outcome of continuing pregnancy, and personal factors are prominent in people's reasons for continuing a pregnancy after considering abortion regardless of policy climate, we found that 14% of those who had decided to continue their pregnancy at follow-up wished they had had an abortion. This suggests that structural-level barriers also play a role in whether people can obtain an abortion. Additionally, even those who got their abortion experienced barriers (such as having to take time off work); some who wanted an abortion may never have progressed far enough in the care-seeking process to identify that as a barrier. As posited by O'Donnell and colleagues (O'Donnell et al., 2018) , when individuals living in restricted access states assess their pregnancy as unacceptable but abortion services do not appear feasible to obtain, they may adjust their emotional orientation towards continuing pregnancy, shifting the continuation of pregnancy to be an acceptable outcome.

Participants identifying as Black had significantly higher odds of still seeking abortion at follow-up, as well as planning to continue their pregnancies, even after controlling for effects of living in a restricted access state and other characteristics. Those identifying as Latinx were also more likely to be still seeking abortion at follow-up, representing prolonged care-seeking. As Boyd et al. (2020) point out, race is not an innate characteristic, but rather should be viewed as an indicator of exposure to racism and other social, environmental, and structural barriers (Boyd et al., 2020) . This analysis is unable to tease out the mechanisms through which racism may be functioning to limit people's ability to obtain abortions. However, these findings suggest that J o u r n a l P r e -p r o o f state-level restrictions may impinge on abortion access differentially by race, in ways that reflect and perpetuate structural racism in health and health care.

There are several limitations to this study. First is possible sampling bias. Internet-based recruitment has been criticized as reaching only a select sample. However, as many as 95-100% of Americans ages 13-49 use the Internet (Pew Research Center, 2019 , 2018 . Multiple studies among abortion patients indicate that they found the abortion clinic through an online search (French et al., 2016; Kavanaugh et al., 2019; Margo et al., 2016) , and even those who may not get an abortion in a clinic still search for abortion information online (Aiken et al., 2018; Jerman et al., 2018; Stephens-Davidowitz, 2018) . Studies demonstrate that the volume of Internet searches for abortion is greater in states with more restrictions and reduced availability of services (Guendelman et al., 2020; Reis and Brownstein, 2010) .

Additionally, differential loss to follow-up may have biased our results. The sample had greater follow-up among those who were older, white, more educated, and did not report difficulty meeting basic needs, suggesting that our estimates may underestimate the effects of barriers on abortion or pregnancy outcomes. Additionally, our multivariable models did not control for other state-level factors that could be relevant to obtaining an abortion, including cultural and social norms. Yet, the state access ratings may be a proxy for the larger acceptability of abortion in the state and there was no differential follow-up by state access rating or state Medicaid coverage.

To date, most studies of barriers to abortion have been conducted among abortion patients; that is, those who were able to make it to the clinic. This paper used an innovative recruitment method and study design to examine barriers to abortion care, representing one of the first attempts to survey pregnant individuals at the point of considering an abortion, well before reaching a clinic. This paper responds to a recent call for more rigorous research on the extent to which women seeking abortion are prevented from obtaining one, which can be considered an indicator of reproductive autonomy (Potter et al., 2019) . Another strength is that the study included a large national sample including individuals from all 50 states. Despite online recruitment, the observed follow-up rate (69%) was consistent with other clinic-based studies (Roberts et al., 2016; Upadhyay et al., 2019a; Weitz et al., 2013) .

State-level abortion policies, including lack of state Medicaid coverage of abortion, are associated with prolonged abortion seeking and not having an abortion at all among this population who were considering abortion and searching online for abortion care. Medicaid coverage of abortion in particular is critical to ensuring that all people who want them are able to obtain abortions. . Percent of participants reporting having to gather the money for travel expenses or to pay for the abortion was a barrier, by state Medicaid coverage of abortion; ***p<0.001 

Motivations and Experiences of People Seeking Medication Abortion Online in the United States

Experiences of women who travel for abortion: A mixed methods systematic review

The effect of facility characteristics on patient safety, patient experience, and service availability for procedures in non-hospital-affiliated outpatient settings: A systematic review

On Racism: A New Standard For Publishing On Racial Health Inequities

Obstacle Course : The Everyday Struggle to Get an Abortion in America

The effects of short-term variation in abortion funding on pregnancy outcomes

The undue burden of paying for abortion: An exploration of abortion fund cases

Influence of clinician referral on Nebraska women's decision-to-abortion time

Women's experiences seeking abortion care shortly after the closure of clinics due to a restrictive law in Texas

Impact of Clinic Closures on Women Obtaining Abortion Services After Implementation of a Restrictive Law in Texas

Change in abortion services after implementation of a restrictive law in Texas

Shining the light on abortion: Drivers of online abortion searches across the United States

Last Five Years Account for More Than One-quarter of All Abortion Restrictions Enacted Since Roe

Restrictions on Medicaid Funding for Abortions: A Literature Review

Barriers to Abortion Care and Their Consequences For Patients Traveling for Services: Qualitative Findings from Two States

Characteristics of U.S. Abortion Patients in 2014 and Changes Since

What are people looking for when they Google "selfabortion"?

At What Cost? Payment for Abortion Care by U.S. Women. Women's Health Issues

It's not something you talk about really": information barriers encountered by women who travel long distances for abortion care

Women's Pathways to Abortion Care in South Carolina: A Qualitative Study of Obstacles and Supports

State Abortion Policy Landscape: From Hostile to Supportive

Radical Attempts to Ban Abortion Dominate State Policy Trends in the First Quarter of

The Safety and Quality of Abortion Care in the United States (Consensus Study Report)

User Manual -Decisional Conflict Scale

Validation of a Decisional Conflict Scale

I wouldn't even know where to start": unwanted pregnancy and abortion decision-making in Central Appalachia

Demographics of Internet and Home Broadband Usage in the United States Fact Sheet

Challenging unintended pregnancy as an indicator of reproductive autonomy

Measuring decisional certainty among women seeking abortion

Measuring the impact of health policies using Internet search patterns: the case of abortion

Do 72-Hour Waiting Periods and Two-Visit Requirements for Abortion Affect Women's Certainty? A Prospective Cohort Study

Complex situations: Economic insecurity, mental health, and substance use among pregnant women who consider -but do not have -abortions

Out-of-Pocket Costs and Insurance Coverage of Abortion in the United States

Estimating the proportion of Medicaid-eligible pregnant women in Louisiana who do not get abortions when Medicaid does not cover abortion

Utah's 72-Hour Waiting Period for Abortion: Experiences Among a Clinic-Based Sample of Women

Association of Facility Type With Procedural-Related Morbidities and Adverse Events Among Patients Undergoing Induced Abortions

The Hyde Amendment and Coverage of Abortion Services

Opinion | The Return of the D.I.Y. Abortion. The New York Times

Intended pregnancy after receiving vs. being denied a wanted abortion

Admitting privileges and hospital-based care after presenting for abortion: A retrospective case series

Using Google Ads to recruit and retain a cohort considering abortion in the United States

Evaluating the impact of a mandatory pre-abortion ultrasound viewing law: A mixed methods study

Denial of abortion because of provider gestational age limits in the United States

Safety of Aspiration Abortion Performed by Nurse Practitioners, Certified Nurse Midwives, and Physician Assistants Under a California Legal Waiver

Experiences Accessing Abortion Care in Alabama among Women Traveling for Services

Abortion as a catastrophic health expenditure in the United States. Women's Health Issues

Figure 1: NARAL Pro-Choice America State Access Ratings 2017-2018 by state* *Protected Access states

West Virginia No state Medicaid coverage of abortion: Alabama

We greatly appreciate the constructive and critical comments from Professor Samantha Baron at the Open University, UK.J o u r n a l P r e -p r o o f