Postabortion contraceptive uptake among vulnerable individuals in a New York City public hospital

Introduction

Across the USA, nearly 40% of pregnancies that occurred in 2015 were either unwanted or occurred sooner than they were desired, and in New York state this estimate rose to 44% in 2017.1 2 Nationally, over 90% of abortions in 2015 were for these pregnancies that were either unwanted or occurred too soon.1 Over 40% of all individuals who underwent abortions in 2018 in the USA had at least one prior termination, a proportion that has remained stable since the early 1990s.3 4 Following induced abortion, ovulation resumes within 3 weeks on average but can occur in as soon as 8 days.5 More than half of people will resume sexual activity in the immediate postabortion period.6 Prompt initiation of contraception after abortion leads to lower rates of unplanned pregnancy and subsequent abortion.7–9 Because the majority of patients seeking abortion do not desire pregnancy in the immediate future, contraceptive counselling and method availability at the time of abortion create an opportunity to empower patients to delay or avoid recurrent pregnancy in alignment with their reproductive goals, while respecting their right to decline or postpone such care.10

Postabortion contraceptive decision-making is complex and multifactorial. A long history of coercive family planning policies in the USA and ongoing race-based discrimination when accessing services has fostered mistrust of the healthcare system and negative attitudes towards contraception, particularly among members of marginalised groups such as racial and ethnic minorities.11 Patients are generally open to discussion of contraceptive options and many actively seek a method around the time of abortion.12 Others may not want contraception or related counselling at the time of abortion, and some may experience or perceive coercion when the topic is raised at this vulnerable time.12 13

While many studies describe sociodemographic and cultural influences on general contraceptive counselling and uptake, there is less information available on uptake in the post-abortion setting. US-based studies often include underinsured patients and find that cost limitations, insurance coverage and method availability impact contraceptive selection and uptake, particularly of long-acting reversible contraception (LARC).12 14 15 Langston et al found that availability of no-cost LARC immediately following abortion was associated with a 43% reduction in the number of patients with repeat abortions within 1 year.7

New York City Health+Hospitals/Bellevue is the oldest public hospital in the USA and serves as a teaching hospital as well as a tertiary referral centre for abortion patients with risk factors or medical conditions that require hospital-based care. As a safety net hospital, Bellevue cares for all patients regardless of ability to pay and serves a disproportionate number of individuals who have experienced systemic racism. Most patients live on low incomes and are publicly insured or uninsured. The hospital includes the cost of intrauterine devices (IUDs) and etonogestrel subdermal contraceptive implants when applying an income-based fee scale to uninsured patients, who may receive all care at no cost depending on their income. Oral contraceptive pills are available at a nominal cost (US$0–US$20 depending on income) or as a covered benefit at the hospital’s pharmacy. Our centre has eliminated most cost-related barriers and uniquely positions us to explore other factors related to contraceptive decision-making among the most vulnerable individuals seeking abortion in New York. This descriptive study characterises demographics, distance travelled for care, clinical factors and contraceptive uptake following medication and procedural abortion at our safety-net public hospital.

Materials and methods

Results

During the study period (April 2019–June 2021), 672 patients received medication or procedural abortions and 638 met inclusion criteria (figure 1). Most were English-speaking, publicly insured, parous and had other race or black race listed in the medical record (table 1). More than half reported not having had a prior abortion. Most abortions were procedural and occurred at a gestational duration of ≥14 weeks (table 1). 165 patients (23%) were referred for hospital-based care for such reasons as anaemia, risk of haemorrhage or risk of complications with anaesthesia. Patients lived a median of 9.05 miles away from the hospital (IQR=6.2–13.1 miles), requiring a median of 36 min travel by car (IQR=27–49 min) or 54 min by public transportation (IQR=43–69 min).

The latent class analysis identified two groups (table 2). The BIC for two classes was 11 295.72 and for three classes was 11 302.85. Members of latent class 1 were more commonly Hispanic, spoke Spanish as a preferred language, frequently had Medicaid insurance and often lived within 10 miles of the hospital. The majority were parous, did not have prior abortions and most often had medication abortions. Members of latent class 2 were mostly non-Hispanic Black, spoke English as their preferred language, had commercial insurance more frequently than class 1, lived farther away from the hospital and commonly underwent procedural abortion ≥14 weeks. The estimated class population shares were 31.7% for class 1 and 68.3% for class 2. Members of class 1 most often chose the subdermal implant or levonorgestrel IUD (LNG IUD) (table 2). Members of class 2 most often chose the LNG IUD or declined contraception. Those in class 1 had a higher chance of choosing the subdermal implant and pills and a lower chance of choosing a barrier method or no method than those in class 2 (table 2). A multivariate logistic regression analysis of demographic characteristics and chosen contraceptive methods was performed but deemed less relevant (data not shown).

A majority of patients chose to initiate contraception during abortion care (509/638, 79.8%). Contraceptive uptake rates were highest among Spanish-speaking patients (115/131, 87.8%), Hispanic/Latinx patients (187/217, 86.2%), those with commercial health insurance (89/108, 82.4%), and those ≥30 years old (246/302, 81.5%). Non-initiation occurred most frequently among individuals who did not have insurance/self-paid (24/72, 33.3%), spoke a language other than English or Spanish (8/27, 29.6%), or had a procedural abortion <14 weeks (38/171, 22.2%). Subgroup analysis did not reveal any statistically significant difference in likelihood of contraceptive choice before and after the beginning of the COVID-19 pandemic (online supplemental Appendix Table 1).

Individuals who initiated contraception most often chose the LNG IUD (134/509, 26.3%, figure 2). An additional 48/509 (9.4%) and 99/509 (19.4%) chose the copper IUD and subdermal implant, respectively, for a total of 281/509 (55.2%) selecting a long-acting method. The subdermal implant was most often chosen among individuals whose ethnicity was listed as Hispanic (29/217, 13.3%) (figure 3A). Patients who had a medication abortion most often chose oral contraceptive pills (51/244, 20.9%). Those who had a procedural abortion <14 weeks most often chose no method (38/171, 22.2%), while those ≥14 weeks most often chose the LNG IUD (64/223, 28.6%) (figure 3B).

In our study population, the proportion of individuals initiating IUDs or the subdermal implant (LARC devices) was highest among individuals undergoing procedural abortion ≥14 weeks (129/223, 57.8%), those whose primary language was Spanish (71/131, 54.1%) and those with reported Hispanic ethnicity (107/217, 49.3%). LARC uptake was lowest among patients who were uninsured or self-pay (22/72, 30.5%), those whose primary language was not English or Spanish (6/27, 22.2%), and those undergoing medication abortion (76/244, 31.1%). Among 348 individuals who lived <10 miles from the hospital, 132 (42.8%) initiated LARC and 66 (19.0%) declined a contraceptive method (figure 3C). Among 290 who lived ≥10 miles from the hospital, 149 (45.5%) initiated LARC and 63 (21.7%) declined a contraceptive method (figure 3C). Of 72 individuals who did not have insurance or self-paid for services, 14 (19.4%) chose oral contraceptive pills, and 10 (13.9%) chose the etonogestrel implant (figure 3D).

Discussion

Most patients seeking abortion at our safety net hospital initiated a contraceptive method, with over 40% initiating an IUD or subdermal implant and the LNG IUD most commonly chosen overall. Latent class analysis identified two groups, each with a similar pattern of contraceptive preference, that differed significantly in demographic characteristics. Latent group 1, which was largely Hispanic and often had medication abortions and Medicaid (public) insurance, was more likely to choose the subdermal implant or oral contraceptive pills. Latent group 2, which was largely non-Hispanic, had procedural abortions and was more likely to have commercial insurance and was more likely to choose the LNG IUD or decline contraception. Overall, uptake was high among Hispanic/Latinx and Spanish-speaking patients, who most commonly chose the etonogestrel subdermal implant. In contrast, individuals who spoke a language other than English or Spanish had lower uptake rates of any contraceptive method. It is not clear whether rates of unintended pregnancy differ in this subgroup, or whether this reflects individual preferences or an unmet need in this population. In our study population, those who did not have insurance or who self-paid for care most often declined contraception. Counselling may need to incorporate financial and logistical considerations among this subgroup, while bearing in mind that method preferences may vary independently of cost. Finally, patients having procedural abortions at ≥14 weeks were more likely to initiate LARC methods than user-dependent methods, while those having a medication abortion often chose oral contraceptive pills, depot medroxyprogesterone acetate injections and barrier methods to LARC methods.

A possible explanation is that individuals choosing a medication abortion have fewer in-person appointments, while those having a procedural abortion later in pregnancy are already having sedation, lowering the potential barrier of pain associated with LARC placement. Other factors external to individual preferences may also play a role, including clinician counselling and underlying biases, as well as system-specific clinical protocols, time constraints on counselling and appointment scheduling.

Among our study population, distance travelled to access abortion care and COVID-19-related concerns appeared to have little impact on the contraceptive method chosen. In the wake of increasing restrictions on abortion in the USA since 2022, individuals are travelling longer distances for this care overall.16–18 We chose a 10 mile distance from the hospital as a cut-off because only 18% of Americans live more than 10 miles from their nearest hospital, suggesting that this threshold demarcates a population with reduced geographic access to care.19 Although patients in our cohort may have had closer alternatives, cost of care rather than time may have driven their decision to travel to our centre. We felt that this threshold represented a substantial travel burden in the context of public transportation availability in New York City. Although our study period predated many state-specific abortion-related restrictions, those who anticipate or experience barriers to accessing abortion care may be motivated to avoid unintended pregnancy. However, personal considerations and other factors may have more of an immediate impact on method choice than distance travelled.

Our study examined contraceptive choice after abortion in a population with a broad method mix available free of charge, regardless of insurance status. In settings where contraceptive methods are not subsidised, insurance status consistently plays a role in contraceptive uptake and even in method selection, with fewer self-pay patients initiating LARC despite their initial preference.12 14 Our findings were consistent with Biggs et al,20 who also found that individuals with insurance more often initiated a method than those without. In their population, 75% used insurance to pay for abortion and most (74%) chose low efficacy methods (barrier methods, emergency contraceptive pills) or medium efficacy methods (combined hormonal contraception, injections). In Roe et al’s study of contraceptive uptake in a setting where LARC was readily available at the time of abortion, nearly 80% had public or private insurance and about 25% initiated an IUD or implant.21 At our centre, patients who cannot provide supporting documentation to obtain state-funded insurance (which covers prenatal care or abortion/contraception at the time of abortion) pay one fee for all care including contraception according to an income-based sliding scale. Many pay nothing for their care given their income. More than 40% of our cohort chose LARC, which was available to all but could also reflect counselling practices and individual preferences. Even among individuals without insurance or those who self-paid for care, nearly 30% chose an IUD or implant.

LARC uptake rates may relate to contradicting desires to avoid future visits for postabortion and contraception care, and/or concern about difficulties in access to effective contraception. Fang et al found slightly higher rates of postabortion LARC uptake among patients who travel longer distances.22 Upadhyay et al showed that patients who live farther away from the site of their abortion care more often sought any necessary follow-up care via the emergency department rather than the outpatient setting, which may reflect a lack of access to or mistrust towards decentralised reproductive health clinicians, especially in the postabortion period.23 Emergency department utilisation may also impede access to LARC follow-up care for patients who live farther from the site of their abortion.

Periabortion contraceptive provision represents an opportunity to reduce the risk of future unintended pregnancies, especially for individuals who experience difficulty accessing abortion care due to insurance status, limited English proficiency or travel. Differences in contraceptive uptake rates between racial and ethnic groups represent an opportunity for clinicians to consider their biases while counselling. This study highlights the benefit of healthcare policies and funding initiatives that allow equitable contraceptive access regardless of ability to pay.

Our study findings indicate that contraceptive uptake is high among abortion-seeking individuals when cost is not a barrier. There is a clear need for further research to understand how other factors influence contraceptive decision-making and uptake around the time of abortion, such as patient–provider communication patterns, cultural influences and patient perceptions about contraception. Further studies may also assess longer-term continuation and satisfaction with methods initiated around the time of abortion. We had planned to assess method continuation through 6 months after abortion, but our chart review found that most patients did not return to our clinic for any contraceptive or gynaecological care. Patients may have had a number of reasons to prefer a different location for this care.

From a policy perspective, our results support the need for no-cost contraceptive programmes, given that those who were uninsured or self-paid most often declined contraception. Our study population demonstrated a relatively high contraceptive initiation rate overall, as well as a high LARC-initiation rate, particularly when compared with other populations that had a greater proportion of uninsured patients.21 Our high uptake rate of LARCs in a no-or-low-cost environment reinforces the findings of Goyal et al, who found that, despite high reported interest in LARC methods among all patients, those who qualified for a no-cost-LARC programme initiated them after abortion significantly more often than those who did not qualify for the programme.24

Our study population comprised a demographically diverse group who had access to a variety of contraceptive methods without cost as a barrier. We examined non-financial factors and contraceptive preferences at the time of abortion in a safety net hospital where all individuals may receive care, regardless of ability to pay. As a referral centre, our patient population comprised many people with complex social and medical needs reflecting many of the most vulnerable in our community.

As a retrospective chart review, we were limited in the ability to directly assess factors affecting contraceptive decision-making. Likewise, motivations behind method selection, reasons for non-initiation and long-term method continuation could not be evaluated. Within the context of a clinical encounter, asking patients why they decline contraception could be coercive even if intended for systematic data collection at a later time. Our findings may not be generalisable to other regions or healthcare centres, particularly those without cost-elimination measures or those with different patient demographics.

There was a high proportion of our study population with other race (45%) and no ethnicity (24%) recorded in the medical record. We subsequently learnt that race and ethnicity are most often assigned by administrative personnel or clerks at the time of chart creation or appointment check-in at our site. When confronted with a limited list and asked to pick a single option that most closely matches a patient’s identity, they may feel limited by the options, feel that none apply, or inadvertently conflate race and ethnicity. They may be uncomfortable asking the patient for a response, feel that they are unable to judge, or believe that it is not their decision to make and therefore decline to provide a more specific answer than ‘other’. These challenges may drive racial and ethnic classification in our hospital’s electronic medical record and may render our analyses based on these factors less useful. Additionally, our study period (April 2019 to June 2021) does not capture long term trends, including those influenced by US policy changes since 2022 which imposed widespread regional restrictions on abortion access. Data during this period may also have been influenced by external factors such as the COVID-19 pandemic.

Future directions for research could include multicentre mixed-method studies using interviews and surveys to qualitatively assess both patient and provider perspectives on postabortion contraceptive preferences. Individuals could provide insight into the factors they consider when making contraceptive decisions, and whether an unwanted pregnancy changes these considerations. Providers could discuss counselling techniques, implicit biases and systems limitations that may affect their recommendations. Clinical trials could explore the utility of patient decision-making tools regarding postabortion contraception.

Conclusions

This study demonstrates the contraceptive choices among individuals seeking abortion at an urban safety net tertiary care centre. By mitigating financial barriers, we highlight factors that may influence contraceptive uptake, including language concordance, proximity to care, race, ethnicity and insurance status. Most of our study population initiated contraception, and differences in rates of LARC and overall method uptake by demographic factors underscore opportunities for tailored approaches. These findings highlight the importance of addressing implicit bias to provide non-coercive counselling that upholds autonomy while also promoting equitable, accessible contraception. Clinical care teams can benefit from training in the dynamics of respectful and plural contraception, especially in a multicultural context. Our study offers insight into potential clinical and policy improvements that can reduce unintended pregnancy rates in at-risk populations.