Building upon the efforts of feminist movements and other country contexts (10, 23) the U.S. can move toward improved surveillance capacity by adopting a definition that harmonizes with existing ones, allowing us to move toward global surveillance capacity. Femicide often is defined as gender-related killing of women and girls (1, 23) and is considered an extreme violation of a woman's right to self-determination (9), depriving her of fundamental human rights to life and bodily integrity, as protected by international law (24). While femicide takes multiple forms, the phenomenon often is grouped into: (1) intimate femicide—femicides perpetrated by a current or former intimate partner and (2) non-intimate femicide–familial femicide, human-trafficking-related femicide, sexual femicide, and crime-related femicide among others (1, 25). Including and clearly defining multiple forms of femicide in a legal definition is “essential to give visibility to the many forms of gendered killings” (10).

We call on the U.S. to follow the example of the Inter-American Model Law on the Prevention, Punishment and Eradication of the Gender-Related Killing of Women and Girls (Femicide/Feminicide) (19) and reform the penal code to include femicide as a form of aggravated homicide as has been done in Argentina, Brazil, and Uruguay (10). Modifications to U.S. criminal statues including a separate category for femicide and clear definitions of its multiple forms would facilitate classification of the gender-related killing of women and girls as femicides, setting critical groundwork for improved surveillance.

To catalog and analyze femicides under the new proposed penal code, existing data structures also need to be updated. Of note, the dynamics of female homicides differ from male homicides (13, 26), including that most female homicides take place in the private sphere (26, 27). As such, contextual information about each homicide is critical. Following guidance from the UN Office on Drugs and Crime on a statistical framework for measuring femicide (28) and research on male-perpetrated female homicides in Canada (23), we recommend that sex/gender-related motives/indicators (SGRMIs) be cataloged and assessed to ascertain whether the killing was a femicide. SGRMIs are characteristics that indicate whether the homicide was “rooted in perpetrators' misogynistic attitudes” (23) and can include factors such as current or past intimate relationship with the perpetrator, familial relationship, perpetrator history of IPV, evidence of sexual violence accompanying the killing, victim experienced human trafficking or involvement in sex work, bodily mutilation and/or public exposure, and evidence of the killing being motivated by hatred of women (23, 26, 28). Accurate and timely collection of the following additional contextual data is necessary to determine the gender-related nature of the killing: gender identity and sexual orientation of the victim, pregnancy status of the victim, perpetrator's history of restraining orders, economic activity status of the victim and perpetrator, and gender-related motive for the killing (28). Currently, the U.S. does not have a surveillance system that collects comprehensive information in these categories for all murder cases. The establishment of a review board, much like the domestic homicide reviews in the UK (20, 22), would be required to collect such data and determine whether a femicide has occurred.

Data on deaths in the U.S. exist under the aegis of public health and criminal justice surveillance systems. In the former, death due to assault can be tracked using vital statistics data from the U.S. Centers for Disease Control and Prevention (CDC) WONDER database. While violent female deaths/homicides are identifiable in these data, they provide limited information about the perpetrator (29). Criminal justice surveillance historically took place through the Uniform Crime Report (UCR), where the Supplementary Homicide Reports included contextual data. In 2021, the UCR was replaced with the National Incident-Based Reporting System (NIBRS), however only 66% of police agencies reported crimes to the new system in 2022 (17, 30), which is similar to the problem with the prior system (12).

The National Violent Death Reporting System (NVDRS), hosted by the CDC's National Center for Injury Prevention and Control, began tracking violent deaths in the U.S. in 2003. This surveillance system compiles facts from death certificates, coroner/medical examiner reports, law enforcement reports, and toxicology reports into one database (31). Where available, the NVDRS includes contextual information about the murder including the relationship between the victim and perpetrator (29, 32). Currently, the NVDRS collects data from 48 states, District of Columbia, and Puerto Rico. While the NVDRS provides the necessary linkages between criminal justice and public health data, the reporting of SGRMIs is not mandatory and known community-level correlates of non-lethal forms of VAW—such as neighborhood poverty, residential instability, and gender inequality (33–35)—are not tracked.

Figure 1 compares the tracked murders of females over time (1980–2020) using data from the CDC WONDER database and the FBI'S UCR. Also included are NVDRS data beginning in 2018, when at least 40 states were included in data collection. The trend lines indicate that the public health system (WONDER) consistently captures more female homicides than does the crime tracking system (UCR). While the crude rate of reported female homicides has decreased since the 1980s, an uptick in female homicides appeared in 2020 during the COVID-19 pandemic, which corroborates other studies (5, 17). Regarding intimate partner homicides, UCR data indicate that intimate partners perpetrate about one third of all female homicides; however, a substantial percentage of female homicides are missing information on victim-perpetrator relationship. Notably, for the years available, NVDRS data indicate higher rates of intimate partner homicides than captured in the UCR data. Finally, the gender identities of the victims are not reported, so some subgroups (e.g., transgender women) may be missing or subsumed into an ascribed category, rendering them invisible.

The NVDRS (black line in Figure 1) is the most comprehensive system to track femicides, and we propose that SGRMIs be included as mandatory fields in homicide reports. We also recommend linking the NVDRS homicide data with census data to detect community-level risk factors. Following the Inter-American Model Law on the Prevention, Punishment and Eradication of the Gender-Related Killing of Women and Girls (Femicide/Feminicide) (19), we further propose that improved NVDRS data be used to establish a femicide observatory that catalogs all cases of killings of women. Several other countries have successfully established femicide observatories; such observatories, run by civil society organizations play an important watchdog role providing contextual and nuanced analysis which supplements national data sources (36).

While an urgent need for improved accuracy in femicide data exists (17, 23, 37), once an adequate surveillance system is in place, questions on risk patterns must be addressed. Crenshaw (38) has emphasized the need to situate women's experiences of violence at the intersection of multiple social hierarchies. Thus, the ability to disaggregate data on femicide by marginalized social identities is critical to identify intersectional risks and impacts and to direct resources to the most vulnerable.

For example, recent research using the NVDRS, despite its limitations, indicates that indigenous women have higher rates of homicide victimization than all other ethnic groups (39). This finding prompted an executive order to address the crisis of missing and murdered Indigenous peoples (40). Black women in America also face a greater risk of being murdered (39, 41), particularly during pregnancy (42). The higher rates of homicide for Indigenous and Black women indicate that the murder of women of color takes place at the intersections of racism and sexism (43). Concerning country of origin, research suggests higher rates of homicide victimization among U.S.-born individuals; however, certain foreign-born groups, including those from Honduras, El Salvador, and Jamaica, have higher than average homicide victimization (44). Notably, these countries have high rates of homicide amidst complex historical legacies of colonialism, slavery, civil conflict, and weak governance (45). Data from Europe indicate that citizenship status is a risk factor for female homicide (46). To our knowledge, female homicide victimization by class or income has not been examined with NVDRS data. Finally, as gender identity and sexual orientation are not included in the aforementioned public surveillance data, the intersectional vulnerabilities of LGBTQ+ people to identity-related homicide are unknown at a national scale. The ability to disaggregate female homicides by other marginalized identities may inform more refined definitions of identity-motivated deaths in the penal code.