Who counts? The invisibility of mothers as victims of femicide

Abstract

This article focuses on the important and persistent phenomenon of women killed by their sons. We argue that parricide (the killing of parents) is a gendered form of violence, given that women are disproportionately represented as victims compared to other forms of violence (aside from domestic homicide by current or ex partners) and that son-mother killings are a form of femicide that is often hidden. Not only do they fall under literal definitions of femicide in that they involve women being killed by men, but they also, we contend, fall under motivation-driven definitions as the killing of women by men because they are women and an institutional state failure to protect them as women. Drawing upon analysis of Homicide Index data and 57 case studies of parricide in the United Kingdom, we show that in many cases women are killed by their adult-aged mentally ill sons, within a broader context of ‘parental proximity’, maternal caregiving and intersectional invisibility, which ultimately renders them vulnerable to fatal violence.

Introduction

The killing of parents by their children, referred to as ‘parricide’, is a gendered form of homicide in which women, like in other forms of domestic homicide, are disproportionately represented (compared to non-domestic violence and homicide), and the vast majority of perpetrators are male. The killing of mothers (matricide) has hitherto fallen under the radar of gender-based violence and abuse research and is not readily identified as a form of femicide. However, nascent UK-based literature on domestic homicide and femicide has begun to acknowledge the problem of mothers being killed by sons and the preponderance of mental illness among son-mother killings.

In this article, we deconstruct the notion of ‘femicide’ and contend that matricide represents a hidden form of femicide that requires recognising and researching as such. Our argument draws initially upon femicide literature, which construes it as the political killing of women by men because they are women, reflecting the impact of the gendered stereotypes, norms and expectations surrounding ‘femininity’, as well as the misogyny experienced by women globally (defined as ‘hatred or dislike of, or prejudice against women’ (Oxford English Dictionary, 2020). We also consider the more politically charged concept of ‘feminicide’, which goes further in holding states to account for their failure to act to protect women and girls at risk of femicide and to punish men who kill women. We then present existing UK-based research on parricide (including our own analysis of Homicide Index (HI) data for 2003–2016) and matricide (including our own contextual analysis of cases from a single UK police force), which indicates that not only does the killing of mothers by sons fit in with narrow definitions of femicide, but it also accords with broader conceptions of feminicide rooted in structural invisibility and state failure to protect.

Constructions of femicide

The concept of ‘femicide’ emerged from the feminist work of Diana Russell, who first coined the phrase in 1976 (Radford and Russell, 1992, cited in Dawson and Carrigan, 2020: 3). Some 40 years later, however, Weil (2016) notes the relative invisibility of femicide literature within sociology. In its narrowest form, ‘femicide’ refers to the killing of women by men, although it is conventionally interpreted as referring to the killing of women because they are women, as a result of gender norms, stereotypes and expectations, and also conveying the prevailing misogynistic attitudes and behaviours underpinning the killing of women, by men (Macaulay, 2021; Walklate et al., 2020; Weil et al., 2018). Over the past 30 years, a range of definitions of femicide have been adopted across the globe, reflecting the broad set of circumstances within which women are killed by men, which in some geographical areas has prompted a stronger political conceptualisation of ‘feminicide’, implicating state responsibility for lack of action to protect women and girls at risk of femicide and failing to hold perpetrators to account (Macaulay, 2021; Shalhoub-Kevorkian and Daher-Nashif, 2013; Shalhoub-Kevorkian, 2002; Toledo, 2017; Walklate et al., 2020).

In contemporary applications of ‘femicide’ and ‘feminicide’, the role of the state has diminished and the term has become less politicised (Macaulay, 2021; Toledo, 2017), although both ‘femicide’ and ‘feminicide’ still strongly convey the context of misogyny. The United Nations definition of femicide (stipulated in the 2013 Vienna Declaration on Femicide) was deliberately depoliticised, adopting the narrower conceptualisation of femicide, which influenced the definition adopted by the European Institute for Gender Equality (EIGE) in 2017 of femicide as ‘the killing of women and girls on account of their gender’ (Weil et al., 2018: 23). According to the United Nations Office for Drugs and Crime (UNODC, 2019), 87,000 women were killed across the globe in 2017, and 58% of these (n = 50,000) were killed by current or former partners, or another family member. Within Europe, it is estimated that 38% of all women killed in 2017 were killed by a current/former partner or ‘another family member’. The report highlights the disproportionate burden of ‘lethal victimization as a result of gender stereotypes and inequality’ experienced by women across the globe, who are killed ‘because of their role and status as women’ (UNODC, 2019: 11).

There is an ongoing drive by criminologists for violence to be more accurately and effectively measured so that fully informed understandings of types of (gendered) violence can be developed and ultimately, responsive prevention measures put into action (Walby et al., 2017). As Walklate et al. (2020) highlight, counting does not automatically lead to the prevention of femicide, but it does assist in raising its profile and furthering the drive to end gender-based violence and abuse. Concomitantly, there is consensus on the need for improved data on the killing of women, with research and prevention in mind, so that more ‘complete’ data that ‘fully captures the nuances, and contextualises the events leading to the death’ can be accessed (Cullen et al., 2021: 2). This reflects the limitations in administrative data, which are collated for the purpose of governments’ needs (Cullen et al., 2021; Dawson and Carrigan, 2020). This not only refers to official statistics, which often fail to record sufficient levels of detail about the killings of women, and subsume femicide within broader homicide statistics, but also to statutory Domestic Homicide Reviews ¹ (DHRs), which do not always capture the social contexts, intersecting social identities and nuances of femicides, despite being introduced to learn from previous mistakes and prevent domestic homicides. The limitations of administrative data mean that certain social identities dominate over others in official narratives of femicide, due to the lack of intersectional data that represent multiple oppressions (Cullen et al., 2021).

The invisibility of mothers as victims of femicide

Definitions and data on femicide, then, are crucial to developing a nuanced understanding of the social and political context of this form of violence, and this includes attention to detail not only in terms of ‘what counts’, but ‘who counts’? The literature on femicide identifies a number of contexts of femicide, including current/former intimate partner homicide, dowry-related disputes, armed conflict, sexual violence, killings of indigenous women, killings relating to sexual orientation or gender identity and so-called ‘honour’ killings (Shalhoub-Kevorkian and Daher-Nashif, 2013; Walklate et al., 2020; Weil et al., 2018); however, the killing of women by former/current partners accounts for the vast majority (UNODC, 2019), and these killings tend to be the main focus. The term ‘other family members’ is often used as a catch-all category for domestic homicides where the perpetrator is not a current/former partner, which according to UNODC (2019) accounted for 20,000 (23%) of all femicides across Europe in 2017. These killings of women are acknowledged as specifically including those by fathers, brothers, uncles and cousins (e.g. UNODC, 2019; Weil et al., 2018), but the category of ‘sons’ as perpetrators is not readily recognised within the literature on femicide – and the lack of data disaggregation makes it difficult to identify the international scale of son-mother killings.

In the following sections, we discuss existing literature on parricide, illuminating its gendered nature and the importance of exploring matricide, which refers to the killing of mothers (in the vast majority of cases, by sons), through a gendered lens. We demonstrate not only the lack of data specifically addressing parricide but also the lack of recognition of matricide as a distinctive form of parricide and a currently invisible form of femicide. In addition to not being acknowledged within the majority of femicide literature (beyond the conflated category of ‘other family member’), data on mothers killed by their sons are submerged within aggregate homicide data in England and Wales, and the contexts within which these women are killed are difficult to ascertain. Since 2011, there has been a statutory duty on local authorities to produce and publish DHRs on all domestic homicides, including parricides – but they are not always completed for parricide cases; DHRs are often difficult to locate and access (despite being public documents), and there is inconsistency in the level of quality within the reviews on the antecedents leading to the killings, particularly in relation to the perpetrator motivation. In the case of mothers killed by their sons, which appears to be predominantly directly related to severe mental illness, there is a lack of accessible data enabling this form of femicide to be acknowledged and understood. The experiences of mothers are therefore obscured in the existing data and literature on femicide, and within the much smaller parricide literature there is a lack of recognition of matricide as a distinctive form.

Parricide as a gendered form of violence

Parricide is a relatively rare event, accounting for approximately 4% of all homicide in England and Wales (Bojanic et al., 2020) and 3%–4% of all homicide across the globe (Holt, 2017). There is little criminological literature focusing on parricide, with the largest body of work emanating from the United States, where Heide and colleagues have published extensively over the past 30 years (e.g. Heide, 2013a, 2013b, 2014; Heide and Petee, 2007). In the United Kingdom, there are no official data specifically focusing on parricide, although the Home Office ² collated HI database can be disaggregated by suspect–victim relationship and gender. Until recently, parricide literature was derived almost exclusively from the disciplines of psychology and psychiatry, with most research focusing on (small) clinical and often forensic settings (e.g. Baxter et al., 2001; Dantas et al., 2014; Liettu et al., 2009; Marleau et al., 2003). In 2017, Holt published an historical overview of parricide across England and Wales, drawing upon HI data from 1977 to 2012, and our recent research has examined the contemporary landscape of parricide, analysing HI data for 2003–2016 as well as exploring the contexts of parricide through a case study analysis of 57 parricide cases from one large police force in England (see Miles et al., 2022 for a full discussion), the findings from which are incorporated into our discussion below. In 2020, Bojanic et al. (2020) published findings from a large, psychiatric-focused study drawing together HI, police and mental health service data, which represents the most comprehensive quantitative analysis of parricide in England and Wales to date.

This emerging body of research consistently reports that parricide is a rare but ‘stubborn’ phenomenon, with numbers averaging between 18 and 22 per year (Bojanic et al., 2020; Holt, 2017; Miles, Condry, and Windsor, 2022), even during the period of national homicide decline witnessed across England Wales during 2004 and 2014 (see Miles and Buehler, 2022). Compared to non-domestic homicide, where women account for approximately 27%–30% of all victims (Miles, 2020; ONS, 2021), females are over-represented as victims of parricide: research consistently reports mothers as accounting for around 50% of all parricide victims (Bojanic et al., 2020; Holt, 2017; Miles et al., 2022), and this supports international literature (see, for example, Bourget et al., 2007; Heide and Petee, 2007). Parricide is also considered a gendered form of violence because men account for 85%–90% of perpetrators (Bojanic et al., 2020; Heide, 2013b; Holt, 2017; Miles et al., 2022).

In terms of perpetrator demographics, the average age is around 31 years (Bourget et al., 2007; Holt, 2017; Miles et al., 2022), and the majority are White (84%), unmarried (89%) and living with parents (70%) (Bojanic et al., 2020). The age of parent-victims varies according to sex, with the risk of matricide peaking for women in their 60s or 70s, while the risk of patricide peaks for men in their 50s (Bows, 2019; Heide, 2013b; Holt, 2017). In our recent analysis of HI data, we found that the mean age for matricides (66 years) was higher than that for patricides (61 years) and also found a statistically significant difference (p < .001) in the age of matricide and patricide suspects: 35% of fathers were killed by a child aged 18–25 years, compared to only 15% of mothers (Miles et al., 2022). Taken together, these data indicate that patricide and matricide are distinctive phenomena, and that the contexts of mother killings are different to the contexts of father-killings.

International research examining the characteristics of parricide perpetrators has predominantly derived from the field of psychiatry and drawn upon clinical and forensic samples, reporting that many parricide offenders have severe mental illness diagnoses, including schizophrenia and psychosis (e.g. Bourget et al., 2007; Catanesi et al., 2015; Dantas et al., 2014; Liettu et al., 2009; Marleau et al., 2003). The dominant criminological framework on the contextual nature of parricide is Heide’s (2013b) typology of parricide perpetrators developed in the United States: the severely abused adolescent; the severely mentally ill parricide offender; and the dangerously antisocial parricide offender. In the United Kingdom, the social and cultural context of parricide may be different, and the differential gun control laws mean that the predominant method of parricide is by sharp or blunt instrument, with very few firearm-related parricides: 2% (Bojanic et al., 2020) compared to 54.9% parricides involving firearms in the United States (Heide and Petee, 2007).

Although Holt’s (2017) analysis of historical HI data (which was supported by our analysis of contemporary HI data, see Miles et al., 2022) indicated that approximately one-quarter of parricides (24% all parricides; 35% of matricides and 14% patricides) were recorded as an ‘irrational act’ (used as an indicator of perpetrator mental illness), one-quarter of defendants successfully invoked the defence of ‘diminished responsibility’ and 31% were sentenced to a secure hospital, leading her to conclude that ‘most parricides are not the product of mental illness’ (p.14; original emphasis); other evidence suggests otherwise. As Bojanic et al. (2020) highlight (and as we articulate in detail in Miles et al., 2022), ‘data analysed in this study did not have a sufficient level of detail to explore the role of mental illness in depth; the proxy variable for mental illness was the notion of an “irrational act” with no analysis by diagnosis’ (p. 517). Indeed, Bojanic et al.’s (2020) analysis of HI data (2007–2014) combined with police national computer and mental health service data revealed much higher levels of mental illness among parricide offenders: 67% of perpetrators had previously been diagnosed with a mental illness, one-third of which involved a diagnosis of schizophrenia or other delusional disorder. Their analysis also highlighted that of the 40% of parricide perpetrators classified as ‘severely mentally ill’, 64% killed their mother, 94% were male, a disproportionate number used sharp instruments (71%), 77% were unemployed, 83% were given a hospital order and 97% were diagnosed with schizophrenia or other delusional disorders. These findings clearly support our earlier assertions (above and in Miles et al., 2022) that matricide represents a distinctive type of parricide and that it needs to be recognised and understood as a form of femicide.

Understanding matricide as femicide

To develop a more informed understanding of the contexts of parricide through a gendered lens, we analysed data from 57 cases from one large police force. Case details were initially provided to us by the police and then expanded with more detailed data from media reports, DHRs, Independent Investigation into Mental Health Reviews (IIMHRs), Adult Safeguarding Reviews (ASRs) and Criminal Appeal Court cases. All review reports were published and publicly available on local authority or health trust websites. The process of manually gathering data on parricides illuminated the challenges in accessing contextual data on a phenomenon which remains on the periphery of homicide (and femicide) discourse. No specific datasets on parricide exist, with data subsumed within homicide statistics, reviews into domestic homicides, adult safeguarding reviews or mental health investigations. As highlighted by Cullen et al. (2021), data collected for administrative purposes, even with prevention in mind, are often inconsistent in terms of quality and unable to reflect the multiple intersecting social identities and oppressions experienced by women killed by men, and this was particularly evident in our endeavours to learn more about the contexts within which mothers are killed by their sons. Ultimately, we constructed a detailed dataset for 57 parricide cases and were able to access at least one full review for 21 of these cases, which provided more thorough and detailed case histories (most of these were for parricides since 2011, when DHRs became a statutory requirement). Each case within the database was analysed to determine a range of victim, offender and incident characteristics, and demographic details. The case reviews were coded thematically to determine common patterns. The study received ethical approval from the University of Oxford Central Research Ethics Committee (CUREC).

Of our 57 cases, most were single-parent parricides, and just two were double-parent. Of the 59 victims, 58% (n = 34) were mothers and 42% (n = 25) were fathers, a gender split reflecting a slightly higher than average proportion of matricides compared to national-level analyses. The age of the victims ranged from 43 to 100 years, with a mean age of 67 years. The mean age of matricide victims was 69 years and slightly higher than patricide victims at 66 years. Notably, only 8% (n = 3) of matricide victims were under the age of 50, compared to 20% (n = 5) of patricide victims, supporting the need to consider patricide and matricide as distinctive phenomena, as the risk appears to differ across the life span. Fifty-three (93%) of the 57 parricides were committed by sons, and just 4 were recorded as perpetrated by daughters (7%), 1 of whom was recorded in the police case data as a trans woman.

Our case study analysis identified the predominance of perpetrator mental illness in these parricide cases: 74% (n = 42) of the 57 offenders were recorded as mentally ill, and for 79% of these (n = 33), the mental illness was found to be legally causative of the killing leading to a conviction for manslaughter due to diminished responsibility, rather than murder. In 7 (17%) of the 42 cases involving mental illness, the mental illness was not found to be legally causative, and in 2 cases, the perpetrator committed suicide and so there was no trial (see Miles et al., 2022 for a more detailed analysis of whether mental illness was found to be factually or legally causative by the courts). When broken down by gender, 77% of matricides (n = 26) and 68% of patricides (n = 17) in this sample were committed by a mentally ill child. These findings unequivocally support recent research reporting mental illness as a significant factor in parricide and suggest that rates of mental illness recorded in HI data are a serious underestimate.

The 21 cases for which we were able to access detailed reviews revealed more about the lived experiences of the victims prior to their death and the contextual circumstances leading to their child killing them. In these cases, we identified ‘parental proximity’, referring to a dependent caring relationship from the parent–child or child–parent. In all but one of these cases, the perpetrator had a serious mental illness. There were 19 cases where a high level of support was being provided in the parent–child relationship and only 2 cases with no such relationship recorded. Of these 19 cases, 14 involved parents caring for a mentally ill child and 5 involved an adult child caring for a parent. The role of the parent carers encompassed a range of practical and emotional support including providing a home, financial support, managing mental health symptoms, organising health care, and support with a range of aspects of daily living. In most cases, the adult child lived with the parents, but those who lived independently still depended heavily on their parents for support.

Within the 14 cases where a parent had been the primary carer, 10 were son-to-mother matricides and the mother had been the primary carer, sometimes with assistance from other family members, but often alone and without any support (whereas the four fathers killed in the context of parental care shared that care with the son’s mother). The data within these 10 matricide reviews are extensive and evocative of the lived experiences of mothers caring for severely mentally ill sons. Our findings are broadly in line with a growing body of research which identifies that when women are killed by their children, they are overwhelmingly killed by mentally ill adult-aged sons, for whom they are the primary carer (see, for example, Benbow et al., 2019; Brennan, 2016; Chantler et al., 2020; Long and Harvey, 2020; Montique, 2019; Sharp-Jeffs and Kelly, 2016). Most recently, the 10-year Femicide Census ³ report revealed that 8% (n = 109) of women who were killed by men between 2009 and 2018 were killed by their biological sons; that 27 of these cases involved ‘overkilling’ (a term used to capture extreme and gratuitous violence); and that mental illness played a key role in these son-mother killings: ‘it is commonly the case that sons, often with mental health issues and with little other ability to look after and finance themselves, are put up by their mothers with attendant risks in the face of a lack of appropriate access to care and support’ (Femicide Census: UK Femicides, 2009–2018, 2020: 61). In our matricide cases, eight were killed with a sharp instrument, one by strangulation and one with a blunt instrument, and there was evidence of ‘overkill’ in eight cases.

Histories of mental illness have been identified in matricide cases in a range of international studies. In a review of the international literature on matricide, Heide (2010) identifies methodological problems which make international comparative work difficult, though perpetrator mental illness is a common theme. Four cohort studies from the United States and England of adult male matricide offenders evaluated in forensic or hospitalised settings with sample sizes ranging from 13 to 58 (Campion et al., 1985; Green, 1981; O’Connell, 1963; Singhal and Dutta, 1992) found that schizophrenia was commonly diagnosed among these adult offenders and they were nearly all single males. A study from South Australia found 11 cases of matricide from 1985 to 2004, 10 of which were perpetrated by sons, one of whom committed suicide following the killing. Seven of the remaining 10 were found not guilty by reason of insanity and an eighth offender had charges reduced from murder to manslaughter due to mental impairment (Wick et al., 2008). A more recent Canadian study found that 60% of matricides were perpetrated by an offender who was seriously mentally ill (Dawson and Hill, 2021).

Heide identifies single case reports from a range of countries including Greece, Germany, Finland, Argentina and Japan (Livaditis et al., 2005; Oberdalhoff, 1974; Silberstein, 1998; Sugai, 1999; Vaisanen and Vaisanen, 1983), where offenders were said to have committed acts of violence against their mothers during a mentally disordered state and locate explanations for the matricide in the mother’s behaviour prior to the killing and long-term dysfunctional relationships. Heide inserts a note of caution in generalising from research derived from small samples of offenders hospitalised for psychiatric evaluation which find most offenders suffering from mental illness as this may not be the case for all matricide offenders. We would similarly note caution in generalising from the explanations provided by offenders in psychiatric settings which do not include the perspective of the mother-victim or other family members. It is interesting that Heide’s review (now 12 years old) identifies potential for prevention in matricide cases:

when mothers present in therapy, clinicians need to inquire about the relationships between mothers and their adult children, particularly when the adult children have significant mental health problems and live in the same home with their mothers. Mothers need to be asked specifically if they are afraid of their sons and daughters, and if they fear that their children could harm them physically. Mental health professionals need to inquire directly if there has been violence in the home because many clients are reluctant and embarrassed to provide this information unless specifically asked. (Heide, 2010: 14)

What is lacking, to the best of our knowledge, is any in-depth research, specifically exploring the contexts, nuances and lived experiences of mothers who are killed by their sons. We highlight here three key themes which emerged from our deeper contextual analysis of matricide cases and demonstrate the need for a gendered lens understanding matricide as femicide. Caution should be noted in generalising from only 10 cases. However, this does not necessarily mean that only 10 of the 34 matricides in our case study sample involved this scenario – there were many ‘unknowns’ in our original police dataset, and it is possible that many more matricides occurred within a similar context, but these details had not been recorded and the mothers’ experiences remain hidden. Summaries of two matricide cases illustrate the themes that we then consider:

Case 1: A 76-year-old mother was killed by her 44-year-old son. He was found wandering the streets naked after the killing. He said he thought his mother was a witch and had mutilated her body using various weapons. She was found on a bed in her home where he was also living temporarily. He had been diagnosed with schizophrenia since the age of eighteen but had stopped taking his medication some weeks before the homicide. Major decisions had been made about his treatment, including not consulting his mother about a decision to allow him to stop his medication, and yet she was expected to monitor his condition and report any problems. A plea of manslaughter on the grounds of diminished responsibility was accepted and he was sentenced to a hospital order under S.37 Mental Health Act with a restriction order attached. It appeared that some services knew that he had previously been violent towards his mother. An ambulance was called to the property for another health issue, shortly before the day of the homicide. He was angry and aggressive towards his mother in front of the crew, and she told them that she was very frightened of her son and that he had the potential to physically hurt her. Although they made a referral to adult safeguarding services, this was not then appropriately investigated or responded to.

Case 2: A 45-year-old man killed his 81-year-old mother. The son lived alone, though near to his mother, and along with other family members his mother was providing a high level of care including managing his finances, and mental and physical health care. Reports describe the man as suffering from autism spectrum disorder and a major depressive disorder with psychotic features and he had been detained under the Mental Health Act prior to the killing. However, he was later assessed as being stable enough for unescorted day release from the hospital psychiatric unit, during which time he visited his mother in her home, as he often did. His mother was known to be frightened of her son and had padlocked some of the rooms in her house. At his mother’s home he attacked both his mother and sister with a glass bottle, causing fatal injuries to his mother. He pleaded guilty to manslaughter on the basis of diminished responsibility and was sentenced to an indefinite hospital order.

Intersectional invisibility of mother-victims

As these cases illustrate clearly, the victims as mother-carers were frequently not ‘seen’ by the services treating their sons and by the professionals they interacted with. They were often completely ignored, their needs were not considered and their concerns were not taken seriously. The National Institute for Health and Care Excellence (NICE) guidelines state that carers’ assessments should be offered to those providing care, so additional supports can be considered such as a carer’s allowance, psychological and family interventions, and these assessments were frequently not offered. One review noted a complete lack of information about the family and the relationship between the perpetrator and his mother in his records and that services had made no effort to contact her. Another noted that the family were not informed about the son’s treatment, medication or the decision to grant day release, despite the mother being his main carer. In one case, contact with the family was only made through male relatives and a brother was offered a carer’s assessment even though he did not live in the family home, which the review notes raises a question of conscious or unconscious gender bias on the part of care coordinators. One review insightfully notes that the mother ‘was not seen as a whole person’: her age, physical health, the demands of caring for her son, and her own needs and wishes were not considered.

The victims experienced ‘intersectional invisibility’ (Purdie-Vaughns and Eibach, 2008), marginalised by multiple subordinate identities which together rendered them invisible and overlooked. They experienced intersecting structures of inequality: as women, as mothers of mentally ill sons, middle-aged or elderly, in some cases as Black and minority ethnic, and with their own health conditions and disabilities, which relegated them ‘to a position of acute social invisibility’ (Purdie-Vaughns and Eibach, 2008). This process is informed by entrenched gender discrimination and the overrepresentation of women in domestic violence and homicide, combined with a pervasive societal climate of mother-blame (Caplan, 2013) and a long history in psychiatry of locating explanations for psychiatric conditions in pathological parenting and the mother–child relationship (the idea of the ‘schizophrenogenic mother’, for example, was widely accepted until the 1970s) (Harrington, 2012). This occurs within a wider context of ‘neoliberal responsibilisation’ which feminist political economists have identified as overburdening women with meeting the caring needs of family members as they are required to ‘take up the slack’ of the retraction of state provided services such as health and social care (Bakker, 2007: 546). This is characterised by a ‘progressive detachment of individuals from social networks and supports, while at the same time, responsibility for systemic problems is being downloaded onto the individual’ (Brodie, 2003: 67)

Middle-aged and elderly women experience cultural invisibility, often poorly represented across many domains, and as Arber and Ginn noted some 30 years ago, elderly women and their concerns have been neglected by both sociology and feminism (Arber and Ginn, 1991). The invisibility of older women as victims of intimate partner’s violence has similarly been highlighted, falling between the margins of domestic violence and elder abuse, and not adequately captured by either field (Crockett et al., 2015). The burden of caring for these women on top of managing their own health conditions and disabilities was largely ignored by services.

We were not able to fully interrogate how race and cultural factors contribute to the intersectional invisibility of matricide victims in our dataset. The victims in two cases were recorded by police as Black, but the reviews state that there was no evidence that this had affected how they were treated – further research is needed to explore how discrimination on the basis of race might intersect with service responses to Black and minoritized women. The organisation Imkaan (2019) notes ‘a lack of specialist knowledge and understanding in mainstream organisations, and racism and discrimination both subtle and overt at sector and state level, [which mean that] many BME survivors are met with insufficient support and inadequate service responses’ (p. 6). Race and cultural factors are therefore likely to be important, but more subtle forms of discrimination might be difficult to evidence retrospectively. It was noted in three cases that the mother’s first language was not English which likely contributed to her marginalisation – in one she was described by a service as difficult to understand due to her accent; in another, the mother needed an interpreter and her son took this role.

Failure to understand risk to mothers

Overwhelmingly, in these cases there was a failure to understand the risk to the mother as a carer to her son. This took two main forms: first, a failure to identify or recognise son-to-mother domestic abuse or to probe further when concerning incidents came to light, and second, a failure to recognise the particular risk to the mother-carer from the symptoms of schizophrenia and psychosis. In some cases, there was a clear history of extensive risk, for example, one son had a previous history of attacking his mother, threatening to kill her, threatening the professionals, holding a doctor hostage with a knife and carrying knives and hammers for protection, yet he was allowed to live with his mother. In another case, there was no previous record of domestic abuse, but the mother had disclosed that her son had been damaging her property, which the review notes fits with the government definition of domestic abuse and should have been responded to as such. In one case, a record had been placed on the son’s general practitioner (GP) notes that he should only be seen by male doctors and that no female doctors should see him due to the risk he posed, yet it was deemed appropriate for him to live with his mother without assessing the risk to her. Overall, the reviews note a poor understanding of the dynamics of domestic abuse from clinicians and care staff. In some cases, this might further have been hampered by the mothers’ fears and concerns about reporting violence from their sons and wanting to protect them.

There was a lack of understanding of the potential risk posed by symptoms of psychosis and delusions centring on mothers. Delusional symptoms such as persecutory ideations, threat/control-override symptoms, command hallucinations and hallucinations of threatening content have all been reported as significant predictors of violence and aggression among patients (Carabellese et al., 2014) and in some cases posed particular risk to mothers as carers. Examples of delusions focused on the mother included a son who said he killed his mother because he thought she was a witch; another because she was evil; and another because she had been replaced by someone else (a phenomenon known as Capgras syndrome and previously reported in matricide cases (Carabellese et al., 2014).

When mothers did interact with services, they were often marginalised and not listened to. There were examples of mothers voicing their fears to clinicians (and in some cases, sons voicing fears that they might hurt their mother) but these were not taken seriously, and referrals were not made to specialist domestic violence services. The mothers were often isolated and the ‘last woman standing’ as services withdrew or failed to properly address the son’s needs and sometimes other family members stayed away due to the son’s behaviour. Most of the sons were already very isolated with few or no friends, or other sources of support. Although mothers were marginalised, they were also often simultaneously responsibilised in their son’s treatment, which included an expectation that they would monitor him and report any decline in his condition and ensure he complied with treatment and attended service appointments, in addition to providing a home and financial and other support far beyond what might usually be expected for an adult child. As one review noted, it is highly risky for family members to challenge someone who has previously been abusive and inappropriate for authorities to put them in a position of responsibility for ensuring his needs are met.

A need to consider histories and trajectories of victims’ lives

Finally, our case analysis highlights the importance of considering the histories and trajectories of the lives of matricide victims (and the lives of mother-carers of mentally ill sons and mothers experiencing violence and abuse) and not solely focusing on the ultimately fatal cases and the act of killing itself. There is an important need to join together with research on adult child to parent violence, of which there is very little, other than within the elder abuse field which does not fully delineate the child to parent relationship (in our HI data, 56% of parents were under the age of 65 and would not fit the definition of ‘elder’), and to make mothers visible and their voices heard. There is a pressing need for research that includes the voices of victims of violence from adult children and the voices of surviving family members in fatal cases.

Research on femicide identifies that it often falls at the extreme end of a continuum of abuse and occurs following years of abuse during which women have been living in fear of violence or their lives (Cullen et al., 2021; Walklate et al., 2020; Weil et al., 2018). Shalhoub-Kevorkian (2002) refers to this as ‘the inability to live’: ‘Although victims of femicide are technically alive, they are in a mode of life that they never wanted and completely reject, a model that is perhaps best described as death-on-life’ (p. 581). This ‘everydayness’ captures the ‘slow femicide’ of women living incredibly difficult lives characterised by fear, isolation and serious detrimental consequences to their own physical and mental health (Walklate et al., 2020).

Within the field of domestic violence, there is little known about how violence intersects with perpetrator mental illness and the consequences for carers. Yet the field of psychiatry does have insights into the experiences of caring for someone with a severe mental illness, which provides an important part of the jigsaw in beginning to understand the contexts of matricide. It plots a shift in the treatment of people with mental illnesses, from institutional to community care, with families now assuming the role of primary caregivers (Mcauliffe et al., 2014). Evidence suggests that people with mental illness achieve better outcomes when they are cared for by their families; however, there is also evidence that the role of caring for mentally ill children has a significant negative impact on carers’ physical and mental health. This body of literature refers to families, carers and parents, yet identifies clearly that this care typically falls onto mothers in their 50s or 60s (Onwumere et al., 2018; Nordstrum and Kullgren 2003). They experience a range of challenging behaviour, including withdrawal, verbal aggression and physical violence, leading to significant stress, loss and fear, feeling unable to cope and feeling trapped, as well as experiencing a sense of guilt, with mothers at particular risk of being victimised by their mentally ill child, ‘because they are the last to give up on their children, despite experiencing difficult or violent behaviour’ (Ferriter and Huband, 2003: 553). Furthermore, they report feeling unsupported by professional staff and experiencing difficulties obtaining crucial information about their child’s illness (Ferriter and Huband, 2003). Onwumere et al.’s (2018) systematic review notes the risk of those with psychosis perpetrating violence towards their carer within the home and highlights the need for a greater understanding of the impact of the emotional and physical burden of caring for a child with these conditions. It also found widespread reports of feeling powerless, frustrated, emotionally distressed and traumatised, and fearful of violence – in some cases, believing their life was in danger. This research documents an established risk of violence towards mothers from their mentally ill children, alongside the increasing reliance upon them to provide care. It also provides an insight into their lived experiences and the extreme emotional and physical burden from this role, including an everyday fear of violence.

Conclusion

Forty years ago, a study of 58 male matricide patients in Broadmoor secure hospital, most of whom had diagnoses of schizophrenia, endogenous depression or personality disorders, concluded that ‘many of the matricides could have been averted by closer psychiatric supervision and care’ (Green, 1981: 214), a statement that remains true today. The number of matricides has remained steady in those 40 years and the experiences of mothers killed, or at risk of being killed, have remained hidden. As we have argued, their invisibility is intersectional and matricide needs to be understood as rooted in structures of inequality and particularly the marginalisation of older women, mothers and carers. We further suggest their experiences are obscured by the siloed production of disciplinary knowledge – the fields of psychiatry and domestic violence have much to learn from each other – and the use of overarching categories of ‘other family member’ in the study of domestic homicide and femicide which obscures their particular experiences as mothers. As we have argued, we need to understand the histories of matricide victims and the lived experience of mothers experiencing or at risk of violence. Women killed by their sons are victims of femicide: as women killed by men; as women killed by men because they are women (as mothers, and in their gendered roles as neglected and isolated carers); and because of institutionalised state failure to protect them as vulnerable citizens.