Emergency departments (EDs) are key sites of health-care delivery in the United States (U.S.), with 141.4 million visits in 2014 alone [1], and 55% of these visits are made by women. Although abortion-related complications are rare [2], the surge in legislation aimed at regulating abortion access [3] suggests that complications are common and that abortion is generally unsafe.

Since 2011, state legislatures across the U.S. have passed numerous laws that regulate abortion provision, many requiring abortion providers to obtain local hospital admitting privileges and have transfer agreements with nearby hospitals [3]. These laws are passed under a presumption that they are needed to protect women’s health and safety [4, 5] and that hospitalization as a result of abortion is an occurrence frequent enough to necessitate legislation formalizing the relationship between hospitals, abortion providers, and clinics.

National-level estimates of abortion-related ED visits do not exist. However, data from one state suggests that abortion-related ED visits are rare. In a study using data from California’s Medicaid program, Upadhyay et al. found that 0.03% of abortions were followed by an immediate ambulance transfer to an ED and 2.6% of abortions were followed by an abortion-related ED visit within 6 weeks of the abortion [2]. Another study that examined all medication abortions done by Planned Parenthood in 2009 and 2010 found an ED treatment rate of 0.10%, although medication abortions represented only about 23% of abortions at the time, and this study included only those ED visits that involved treatment [6]. In a study of outcomes of abortion procedures by family physicians in New York and Philadelphia, 0.3% of first-trimester medication and aspiration abortion patients were referred or went to an ED for assessment [7].

There is a paucity of published national data on the incidence and outcomes of ED visits after abortion. In this study, we examine the frequency of abortion-related ED visits, the frequency of major abortion-related incidents, and the characteristics of abortion-related ED visits in the U.S. using a nationally representative sample of ED visits.

Among 42,493,214 unweighted visits among women of reproductive age (15–49) during the study period, 6342 visits had an abortion diagnosis. Among these, 70 visits were determined to be unrelated to abortion based on clinician review and 33 visits were duplicates (all variables were identical other than ID number); these 103 visits were excluded, leaving an analytical sample of 6239 unweighted abortion-related ED visits. We did not include an additional 101 visits with abortion diagnoses for women outside the age range of interest (younger than 15 or older than 49).

The final analytical sample corresponded to 27,941 weighted ED visits for abortion-related reasons among 189,480,685 weighted ED visits among women of reproductive age. Thus abortion-related ED visits represented 0.01% of all ED visits among women of reproductive age during the study period. All subsequent results are weighted.

Three-quarters (75.1%) of abortion-related ED visits resulted in discharge from the ED. Another 21.2% were admitted to the same hospital for abortion-related reasons and 0.8% were admitted for non-abortion-related reasons. For the remaining 3%, the disposition was unknown or they were transferred to another medical facility or left against medical advice. Among all abortion-related ED visits over the 5 years of data (n = 27,941), 15 ended in the patient’s death.

We used published rates of the total number of abortions to calculate the percentage of all abortions seen in an ED that resulted in a major incident. During the 5-year study period, there were an estimated 5,282,500 total abortions in the U.S. [17]. Using this number as the denominator, we estimate the rate of major incidents seen in EDs for abortion in the U.S. is 0.11%, or 108 per 100,000 abortions.

We found that abortion-related ED visits comprised 0.01% of ED visits among women aged 15–49. In other words, 14 of every 100,000 ED visits among women aged 15–49 were for abortion-related reasons. The majority (51%) of these were visits involving observation care only.

These data also allowed us to estimate the national major incident rate after an abortion. The rate of 0.11% (108 per 100,000) abortion patients is slightly higher than the rate of 0.05% found in a study of first trimester abortion patients in California [18], and slightly lower than the rate of 0.23% found in a study of all abortions covered by California’s Medicaid program [2]. While not all abortion-related incidents lead to an ED visit and thus, are not reflected in this estimate, we believe that the vast majority of major incidents (those involving a blood transfusion, surgery, or hospital admission) are reflected here. Those that would be missed from this analysis are cases that skip the ED and are directly admitted to a hospital or an outpatient surgicenter (usually for scheduled hospitalization) and complications arising from the small proportion (4%) of abortions done in hospitals [17] and are then directly admitted. Thus, the major incident rate may be slightly underestimated.

The major incident rate for abortion (0.1%) is lower than the published rates for pregnancy (1.4%) [19], as well as other common procedures such as colonoscopy (0.2%) [20], wisdom tooth removal (1.0%) [21], and tonsillectomy (1.4%) [22]. Abortion care is, thus, safer than many other unregulated outpatient procedures. Additionally, we found 15 deaths between 2009 and 2013, which is slightly lower than the total number of abortion-related deaths reported in 2009–2012, the most recent years available (n = 24) [23].

Notably, the majority of visits involved observation care only, which is consistent with a previous study [2]. Patients experience a range of post-abortion symptoms, including ongoing uterine cramping and bleeding for up to 3 weeks after the abortion. Patients may not realize that this is normal. Some women do not start to bleed until several days after the abortion, while some stop bleeding and then start again. Increased cramping and bleeding could start several days after the abortion. Patients may not be given ample information about what to expect or they may have trouble differentiating normal post-abortion symptoms from signs of a complication. Patient visits to EDs for non-urgent care have the potential to be costly to the health system. Such visits could be due to several reasons and little research has been done on factors that contribute to patients’ decisions to visit an ED after abortion. EDs offer 24-h access to care compared to abortion facilities, which have relatively limited hours and may require an appointment [24]. The long distances that many women across the U.S. must travel to reach an abortion provider may make return visits for follow-up too arduous [25, 26]. Indeed, research found that patients who travel longer distances to reach an abortion provider are more likely to visit an ED for follow-up care or to manage subsequent symptoms and less likely to return to the original abortion provider [27]. We note that patients presenting to the ED were disproportionately of urban residence (91% in the sample compared to 81% of the U.S. population) [28]. This is likely explained in that 91% of abortion patients live in urban areas [29] and 92% of EDs in the sample were in urban areas, attracting mainly local residents who may find them convenient geographically.

Post-abortion visits to the ED may also be driven by stigma, worry, or distrust of abortion providers. A perception that abortion is unsafe [30–32] may lead women to worry about mild symptoms, such as cramping and bleeding, even though they are an expected result of abortion. Such perceptions may stem from abortion portrayals in the media and popular culture. A study of abortion-related storylines in fictional American television shows found a major incident rate of 34% [33], over 34,000% greater than the real-life major incident rate of 0.1% found here. The long-term consequences of these fictional abortions were much more likely to be severe, including frequent depictions of negative mental health sequelae, infertility, and even death.

Women using Medicaid had higher odds of major incidents than those not using Medicaid and lower odds of observation care. In this context, insurance type may be a proxy for socioeconomic status, as women requiring Medicaid are low income and as a result, face a multitude of barriers to accessing health care and are known to have poorer health status, including multiple chronic conditions, than women with private insurance. Women who were self-pay were less likely to have major incidents and more likely to receive observation care only, suggesting that patients without healthcare coverage may not have been given treatments to reduce patient costs.

We found that the pre-existing chronic conditions that have previously been suspected to be associated with major abortion-related incidents were indeed associated with a significantly higher rate of those incidents in this sample. While there is limited previous research on the impact of chronic health conditions on the risk of abortion complications, it is well established that women with chronic conditions are more likely to have pregnancy-related complications [34]. Our findings are consistent with provider guidance that suggests women with multiple chronic medical conditions may be at increased risk [35], but conflicting with previous studies that find that obesity and chronic health conditions confer no increased risk among women having abortions [11–14]. This increased risk might be explained in that women who had overnight inpatient stays (one of the categories of a major incident) were more likely to have their obesity or other chronic diseases documented in their charts than those who did not have an overnight stay.

We found some cases of potential self-induced abortions using such means as poisoning or other methods of self-harm. While self-induced abortions are safe when appropriate dosages of mifepristone and misoprostol or misoprostol alone are used, other methods are hazardous, as evidenced by ED visits. Rates were highest in the South, which is known to have the most barriers to abortion access, including the fewest providers [17, 26], and the most state-level restrictions. States that are hostile to abortion may see more ED visits due to self-induced abortion than non-hostile states, potentially due to stigma, protesters, and other barriers to in-clinic abortion.

This study included a large, nationally representative sample of ED visits, allowing us to draw national- and regional-level conclusions about abortion safety. However, using billing codes to understand the nature of the ED visit can be imprecise and incomplete. The estimates produced here may be conservative if patients did not report having had an abortion due to fear of stigmatization or if relevant diagnosis and procedure codes were not reported or were systematically misreported. The lack of full clinical data to determine abortion relatedness could cause errors. For example, the visits in this study could include cases of miscarriage. Likewise, this study may miss abortion-related incidents that were inaccurately coded as a miscarriage.

These new findings can inform policy debates regarding abortion regulation in the U.S. Regulations on abortion provider or facility relationships to hospitals or EDs should be considered in light of their relative impact on improving women’s health. Because abortion-related ED visits comprise a very small proportion of women’s ED visits, and the rate of major incidents is very low, regulations on abortion are unlikely to have any impact on women’s health outcomes. Many abortion-related ED visits are for observation only and may not be indicated or could be managed at a less costly level of care. Perceptions that abortion is unsafe are not based on evidence.