In its 2022 decision in Dobbs v. Jackson Women’s Health Organization, the U.S. Supreme Court ruled that the Constitution does not guarantee a right to abortion and gave states the ability to restrict abortion access as long as the rationale for those restrictions serves a legitimate state interest. In the wake of that decision, 14 states have completely banned abortion with very few exceptions, such as to prevent the death of the pregnant person (Felix et al., 2024). Another 27 states have abortion bans based on gestational age (Guttmacher Institute, 2024).

Although abortion access has been dramatically curtailed in the past two years, states have been gradually restricting access to abortion since the early 1990s as part of a concerted effort to dismantle federal abortion protections (Roth & Lee, 2023). Some pro-choice scholars and advocates have responded to the imposition of these restrictions by focusing on the potential consequences of denying pregnant individuals access to abortion (Foster, 2021; Greene Foster et al., 2019; Wilkinson & Bernard, 2024). Historically, efforts by scholars to link increased abortion access to decreased child maltreatment rates (Bitler & Zavodny, 2002, 2004; Ney et al., 1993; Seiglie, 2004; Zuravin, 1987) have had limited success due to the many conceptual and methodological issues we address in this paper. Most recently, Adkins and colleagues (2024) concluded that the imposition of Targeted Regulation of Abortion Providers (TRAP) laws by states was associated with an increase in state foster care entries, particularly entries due to inadequate housing. TRAP laws are medically unnecessary, politically motivated regulations imposed on abortion providers that result in substantial barriers to those attempting to access abortions (Mercier et al., 2016).

Without careful contextualization, findings from studies linking reductions in abortion access to increases in child maltreatment could be misinterpreted as evidence that people who become parents after being unable to access abortion care are more likely to maltreat their children. Likewise, research connecting abortion restrictions to increased foster care entries risks implying that children are in foster care because their parent could not obtain an abortion. In reality, child maltreatment and entry into foster care are caused by a multitude of factors. As such, disentangling the relationship between abortion access and either child maltreatment or foster care entries is difficult for both methodological and conceptual reasons and requires a nuanced understanding of child welfare systems (Finkelhor & Jones, 2006; Li & Liu, 2024).

In this paper, we highlight some of the conceptual and methodological challenges that using national child welfare datasets to examine the link between abortion restrictions and either child maltreatment or foster care entries presents. We explore challenges related to (1) the magnitude of the effect of abortion restrictions, (2) operationalization of child maltreatment, and (3) the use of national child welfare datasets to examine links between abortion restrictions and either child maltreatment or foster care entries. We also challenge some of the assumptions underlying this research and argue that researchers using national child welfare datasets to demonstrate a causal relationship between abortion restrictions and either child maltreatment or foster care entries should exercise caution.

Magnitude of Effect: Abortion Restrictions do not Equate to Widespread Forced Birth

The negative effects of limiting reproductive autonomy by restricting access to abortion can be profound (Londoño Tobón et al., 2023). Pregnancy carries far greater health risks than legal abortions, as evidenced by the high maternal mortality and morbidity rates in the U.S. (Fink et al., 2023; Hoyert, 2022). These risks include gestational diabetes, hypertensive disorders of pregnancy, thromboses, anemia, hemorrhage, eclampsia, obstructed labor, and sepsis (Keegan et al., 2023). For every 100,000 live births, there are 23.8 deaths of women (Hoyert, 2022). In comparison, the rate of fatalities from legal abortions is estimated to be 0.43 per 100,000 abortions (Kortsmit et al., 2022). Being denied abortion care may also have negative effects on mental health, such as increased anxiety and lower self-esteem, at least in the short-term outcomes (Biggs et al., 2017). Some pregnant people will seek to terminate their pregnancy by obtaining abortions in unregulated, unsafe settings, ingesting toxic substances or self-inflicting physical injury (Keegan et al., 2023). Not being able to obtain an abortion can adversely affect the economic well-being of pregnant people and their families (Foster, 2021).

Although terminating a pregnancy is far more difficult than it used to be for pregnant people in states with abortion bans, not all pregnant people who want to terminate their pregnancy will be forced to give birth. Some pregnant people in states with abortion bans may choose to terminate their pregnancy by traveling to clinics in states where abortion is legal. To illustrate, more than twice as many people traveled out of their state to obtain an abortion in the first six months of 2023 compared to the last six months of 2022 (Forouzan et al., 2023).

Other pregnant people in states with abortion bans choose to terminate their pregnancy via a medication abortion within or outside the formal healthcare system. Between 2020 and 2022, the number of clinics, both brick-and-mortar and virtual-only, providing telehealth with the option to mail abortion pills to pregnant people rose from 52 to 243, while the number of virtual-only clinics providing abortion medication services rose from 0 to 69 (Upadhyay et al., 2024). In December of 2023, telehealth abortions (i.e., medication abortion provided by a clinician through a remote consultation that results in medications dispensed via mail) accounted for 19% of all abortions nationwide (Norris et al., 2024).

Self-managed medication abortions are more difficult to count because they occur outside of the formal health care system. However, one study found that the number of daily requests received by the nonprofit AidAccess, the sole online telemedicine provider of self-managed medication abortion in the U.S., increased by 158%, from an average 83 per day in the baseline period (September 1, 2021, to May 1, 2022) to an average of 214 per day in the immediate post-Dobbs period (June 24, 2022, to August 31, 2022). Moreover, the states with the largest increases in their weekly request rate were states with abortion bans (Aiken et al., 2022).

In short, although clinics are no longer performing abortion procedures in states with abortion bans, many pregnant people in those states who want to terminate a pregnancy are still able to do so–although sometimes at significant financial and emotional cost. In fact, the number of abortions provided in the formal health care system nationally increased 11% between 2020 and 2023 (Maddow-Zimet & Gibson, 2024). Importantly, not all pregnant people who would choose to terminate their pregnancy are able to access abortion. A 2023 analysis estimated that in the first six months of 2023 roughly one-fifth to one-fourth of people seeking abortions did not receive them due to bans (Dench et al., 2023). However, the likelihood that denying abortions to this group of pregnant people would produce a detectable effect on an outcome like child maltreatment reports or foster care entries–that is driven by a convoluted mix of state and child welfare policies and factors is exceedingly small, especially in the short term.

Complexities of Studying Child Maltreatment

Studying child maltreatment is inherently complex and requires a nuanced understanding of child welfare systems. Researchers planning to study the link between abortion restrictions and children maltreatment need to consider both how they will measure child maltreatment and how they will account for the complex relationship between poverty and neglect.

Measuring Child Maltreatment

We do not know the “true” rate of child maltreatment because not all child maltreatment is reported. Moreover, the national child welfare dataset that is most often used to study child maltreatment (i.e., the National Child Abuse and Neglect Data System [ NCANDS]) only includes alleged child maltreatment reports to which child protective services responded with an investigation or alternative response (Fallon et al., 2010). Thus, child welfare researchers rely on proxy measures, each of which has limitations. For example, using the number of child maltreatment reports investigated is potentially problematic for two reasons. First, decisions about which reports of maltreatment to investigate depend on statutory definitions of maltreatment, which vary across states. For example, some states include exposure to domestic violence in their definition of child maltreatment whereas others do not (LaBrenz et al., 2020). Second, most reports of alleged maltreatment investigated by child protective services are unfounded either because no maltreatment occurred (e.g., a child’s injury was accidental) or because not enough evidence to support the allegation was found. For example, in federal fiscal year 2022, 16% of children who were the subject of an investigation or alternative response were determined to be victims of child maltreatment (U.S. Department of Health & Human Services, Administration for Children and Families, Administration on Children, Youth and Families, Children’s Bureau, 2024).

Another option for measuring child maltreatment is to look at substantiated reports of maltreatment. Substantiated reports are reports for which sufficient evidence exists that maltreatment occurred. While this might seem to be the most straightforward approach to measuring child maltreatment, it is potentially problematic for three reasons. First, statutes defining maltreatment and the level of evidence required for substantiation vary widely across states (Lee & Weigensberg, 2019; Rebbe, 2018). Second, the decision to substantiate a report is influenced by a variety of family, caseworker, and agency factors, and the agency factor may be a function of state policies (Day et al., 2011; Font & Maguire-Jack, 2015; LaBrenz et al., 2020). Third, unsubstantiated cases do not necessarily equate to no maltreatment (Drake, 1996).

Finally, a common proxy for child maltreatment is foster care entries. Using entries into foster care as a proxy for child maltreatment is problematic for several reasons. First, most children who are victims of child maltreatment do not enter foster care. For example, in federal fiscal year 2022, approximately 20% of child victims of maltreatment were removed from their homes within 90 days of the investigation disposition date (U.S. Department of Health & Human Services, Administration for Children and Families, Administration on Children, Youth and Families, Children’s Bureau, 2024). By contrast, 55% of child victims of maltreatment or the families of those victims received some type of post-response service, such as case management, counseling, daycare, mental health or substance abuse treatment, or respite care. Second, children enter foster care for reasons other than maltreatment. One recent study found that between 8% and 35% of children entering foster care did not experience a child abuse or neglect investigation prior to their entry (Drake et al., 2022).

Researchers examining the relationship between abortion restrictions and child maltreatment should understand the limitations of the measures of maltreatment they are using and clearly explain those limitations to both academic and lay audiences. If not, their findings may be misinterpreted and invalid conclusions may be drawn.

Distinguishing among Abuse, Neglect, and Poverty

The argument that abortion restrictions lead to an increase in child maltreatment rests on the assumption that parents who would have chosen to terminate a pregnancy in the absence of abortion restrictions are more likely than other parents to maltreat their children. One problem with this assumption is that child maltreatment includes both acts of commission (i.e., physical, sexual, and emotional abuse) as well as acts of omission (i.e., neglect). We are not aware of any evidence that parents who sought but could not obtain an abortion are more likely to physically, sexually, or emotionally abuse their children than other parents. Consequently, if abortion restrictions lead to an increase in child maltreatment, that increase is most likely to be driven by an increase in neglect.

The vast majority of child protective services investigations involve allegations of neglect (Palmer et al., 2024). Neglect also accounts for the vast majority of substantiated child maltreatment reports. 74% of child maltreatment victims in Federal Fiscal Year 2022 experienced neglect (U.S. Department of Health & Human Services, Administration for Children and Families, Administration on Children, Youth and Families, Children’s Bureau., 2024).

Neglect is typically defined as a failure to provide essential age-appropriate care including food, clothing, shelter, medical care, or supervision (Dubowitz et al., 2022). A majority of states explicitly exclude the financial inability to provide for a child’s care from their definition of neglect (Williams et al., 2018) and the Children’s Bureau’s definition of neglect includes a “poverty exemption.” However, poverty and neglect are highly correlated (Skinner et al., 2023). Poor children are over-represented among victims of maltreatment (Dolan et al., 2011: Sedlack et al., 2010), and the relationship between poverty and neglect is especially strong (Sedlack et al., 2010).

Less clear is whether the relationship between poverty and neglect is causal, and if it is, how poverty contributes to neglect (Berger & Waldfogel, 2011). Poverty may contribute both directly and indirectly to neglect (Pac et al., 2023). Families experiencing poverty may be unable to afford the food, clothing, medical care, housing and childcare their children need, and poverty may increase parental stress, which can adversely impact parenting behaviors. Poverty is also associated with other contributory factors, such as poor mental health, substance misuse, and domestic violence (Skinner et al., 2020), and poor families may experience heightened surveillance and scrutiny due to frequent interactions with social service providers and other mandated reporters (Dettlaff et al., 2020; Merritt, 2021). This may explain why poor children are over-represented among victims of maltreatment (Kim et al., 2018).

From a social justice perspective, the distinction between abuse and neglect is critical to understanding the relationship between abortion restrictions and child maltreatment. If abortion restrictions adversely affect the economic well-being of pregnant people and their families (Foster, 2021), this could potentially lead to an increase in foster care entries, not because children are being abused parents who would otherwise have terminated a pregnancy, but rather, because poverty is often confounded with neglect. Thus, researchers studying the link between abortion restrictions and child maltreatment must include measures of poverty as a mediating variable in any analysis and clearly discuss the role of poverty when interpreting their findings.

Limitations of Research Using National Child Welfare Datasets

Researchers who want to study the link between abortion restrictions and child maltreatment or foster care entries typically use national child welfare datasets because they include data from all, or almost all, states (Yi & Wildeman, 2023). These include the National Child Abuse and Neglect Data System (NCANDS) and the Adoption and Foster Care and Analysis Reporting System (AFCARS). NCANDS is a voluntary system for collecting data from all 50 states, the District of Columbia, and Puerto Rico about child abuse and neglect reports. Each year, states submit a file containing child-specific records for each report of alleged child maltreatment to which child protective services responded with an investigation or alternative response (Waldfogel, 2000). AFCARS is the system to which state and tribal agencies are required to report case-level data on all children in foster care and children who have been adopted. Twice per year, states are required to submit their data in order to receive payments through Title IV-E of the Social Security Act (Yi & Wildeman, 2023).

Two major limitations of both NCANDS and AFCARS are variability in the data across states and missing data (Green et al., 2015; United States General Accounting Office, 2003). For example, the way child maltreatment is defined is not uniform across states. Consider the definition of neglect. Some states define neglect to include educational neglect; others do not include educational neglect in their definition. To take another example, children can be removed from home due to inadequate housing; some states report whether inadequate housing contributed to a child’s removal while others do not.

One recent example illustrates both problems. Jones and colleagues (2024) examined the poverty-related variables in NCNADS and noted that states have vastly different reporting standards. In some years, states may not report any data for poverty-related variables, or the reported data may be unreliable or inconsistent. For instance, in the 2019 NCANDS child data file, the variable that measures public assistance receipt was coded “no” for all the children from Colorado; that same variable was coded “unknown” for all the children from New York, Ohio and Oregon. Different variables may be affected by reporting issues in different years and states may have reporting issues in some years but not in others.

Researchers who use NCANDS or AFCARS data to study the link between abortion restrictions and child maltreatment or foster care entries must take these limitations into account when analyzing their data. In some cases, this may mean excluding states with implausible data values or states missing data for the majority of children. Failure to do so can lead to inaccurate or biased conclusions.

Researchers who use these national datasets to study the link between abortion restrictions and child maltreatment or foster care entries must also be aware of changes in state- and county-level child welfare policies that can influence child maltreatment reporting and/or entries into foster care over time (Russell & Macgill, 2015). Failing to take these changes in child welfare policies into account can result in invalid conclusions about the relationship between child maltreatment or foster care entries and abortion restrictions.

To illustrate how this can happen we cite two examples. Texas enacted some of the country’s most restrictive abortion policies in 2021. If abortion restrictions are driving foster care entries, the number of children entering foster care in Texas should have increased once these restrictions took effect. However, the state experienced a 40% decrease in foster care entries. This decrease was most likely due to recent policy changes. Most notably, in 2021, House Bill 567 narrowed the legal definition of child neglect, from actions or inactions that expose children to a substantial risk of harm to actions or inactions that result in immediate danger (Texas Department of Family and Protective Services, 2024). By contrast, Illinois has no TRAP laws (Guttmacher Institute, 2023b), but the state’s foster care population rose 32% between Federal Fiscal Year 2017 and Federal Fiscal Year 2021 (US Department of Health and Human Services, 2022).

Accounting for all the changes in state- or county-level policies that could explain changes in child maltreatment or foster care entries might not be feasible, particularly given that difference policies change at different times and in different ways across jurisdictions. However, researchers using national datasets to study the link between abortion restrictions and child maltreatment or foster care entries should at least acknowledge their potential impacts.

Child welfare policies are not the only factors that can contribute to changes in child maltreatment reporting or foster care entries. For example, the foster care entry rate is dependent on many macro-level factors, including supports available to families, alternative response or other diversion programs, and, as already noted, changes in the threshold for removing children from their homes (Kahn et al., 2017). Such changes can occur even without a formal change in policy. For example, foster care entries can increase following the death of a child who was not removed from home despite documented allegations of child maltreatment (Yelderman et al., 2022).

Special Cases in Foster Care and Adoption

One child welfare outcome that has received much less attention from researchers investigating the effects of abortion restrictions on child maltreatment or foster care entries is infant abandonment. Abortion restriction could lead to an increase in cases of infant abandonment if pregnant people who would otherwise have terminated their pregnancy are forced to give birth. All 50 states plus the District of Columbia allow parents or, in some cases other specified parties, to relinquish an infant under certain circumstances without threat of prosecution for child abandonment, but the circumstances under which an infant can be legally abandoned vary widely across states (Guttmacher Institute, 2023a). In each of the five most recent reporting years for which AFCARS data are available (FFY 2018 - FFY 2022), 5% of all children entering foster care were abandoned and 1% were relinquished. Importantly, these percentages include children of all ages. Further evidence that the incidence of infant abandonment is very low comes from the National Safe Haven Alliance, which estimates that just over 4,500 were relinquished under safe haven laws between 1999 and early 2024 (National Safe Haven Alliance, 2024). The low incidence of infant abandonments would seem to preclude detecting any effects of abortion restrictions on this outcome.

Another option would be to examine the relationship between abortion restrictions and private infant adoptions. Because relatively few infants in foster care are waiting to be adopted, some prospective adoptive parents work with a private adoption agency to adopt an infant directly from the birth mother rather than adopting a child in foster care. In theory, one could examine when the imposition of abortion restrictions is associated with an increase in the adoption of infants outside the child welfare system. However, no national tracking system captures all private adoptions (Sisson, 2021).

Conclusion

The rapid proliferation of abortion restrictions and outright bans in the wake of the Dobbs decision raises important questions about the impact of these policy changes on child maltreatment. However, linking restricted abortion access to child maltreatment requires researchers to consider the both actual impact of abortion restrictions on the occurrence of abortions and the methodological and conceptual challenges that studying the link between abortion restrictions and child maltreatment presents.

Research has shown that pregnant people who are denied an abortion are more likely to experience poverty, raise their child alone, remain in contact with a violent partner, and report poorer mental health than those who seek and receive an abortion (Foster, 2021). Because these are also risk factors for neglect (Dubowitz et al., 2022), the imposition of abortion restrictions could contribute, indirectly, to an increase in child maltreatment.

At the same time, research on the impact of abortion restrictions has the potential to create false and even harmful narratives about parents who were forced to give birth and the children they were not planning to bear. We are not aware of any evidence that mothers who are forced to give birth due to abortion restrictions are more likely to abuse their “unwanted” children or that abortion restrictions are leading to a growing number of “unwanted” children entering foster care. However, if investigations linking abortion restrictions to child maltreatment or foster care entry rates fail to consider the complexities of child welfare systems and the limitations of national child welfare datasets, spurious relationships may be observed that lead to false and stigmatizing narratives.

As child welfare researchers, we are deeply concerned about how abortion restrictions will further limit not only bodily autonomy but also access to contraception, testing for sexually transmitted infections, and a full range of sexual and reproductive healthcare as clinics that once provided abortion services are forced to close. We argue that investigators seeking to study the impacts that abortion bans or restrictions have on child maltreatment or foster care entries should account for potentially confounding variables in their analysis, acknowledge the limitations of the data they are using, and exercise caution when interpreting their findings. We also argue that researchers should consider using prospective cohort studies or qualitative methodologies to explore what is likely a complex relationship between abortion bans or restrictions and child maltreatment.