Too risky yet not risky enough: the intersecting characteristics, vulnerabilities, harm indicators and guardianship issues associated with seriously harmed missing children

ABSTRACT

The number of missing child reports exceed police investigative capacity, yet some incidents are linked with harm, making effective risk assessment essential for safeguarding. Police data likely underrepresents harm to missing children due to harm being undisclosed, and missing incidents going unreported. A better understanding of harm associated with missing children could help to develop appropriate interventions to reduce missing incidents and prevent harm. This study examined 18 months of published Child Safeguarding Practice Reviews across England – a previously overlooked resource for understanding harm to missing children – to identify the intersecting characteristics, vulnerabilities, harm types, indicators, and issues with formal guardianship (safeguarding by carers, schools, local authorities, police and health professionals). Results revealed that children were missing, vulnerable, harmed, and showed indicators of exploitation in numerous and intersecting ways. Of the missing from home or care (MFHC) cases, all but two mentioned repeat missing incidents, all mentioned missing school, 75% mentioned exclusion, and over half mentioned going missing from both home and care. All MFHC cases mentioned multiple vulnerabilities and multiple harm types. All but one of the MFHC cases mentioned multiple exploitation indicators. This highlights opportunities for guardianship including extending the police Philomena Protocol to schools for children at risk of exclusion. The constraints of formal guardians, when combined with the risks faced by some vulnerable children, can lead to inadequate supervision and support – paradoxically leaving those children simultaneously overexposed to harm yet under-served by protective services.

KEYWORDS:

• Missing children
• harm
• guardianship
• policing

Introduction

Children going missing is an international concern (see Stevenson and Thomas 2018, Fedina et al. 2019, Ferguson et al. 2023), not least given its ties to cross-border child trafficking (Sharp-Jeffs 2016) yet there is relatively little research on missing children with most research emanating from the UK (Ferguson and Sidebottom 2022). Definitions of a missing person vary across countries (Shalev Greene et al. 2022). In England, a missing child is anyone under 18 years of age whose location cannot be determined (College of Policing 2014, 2016). Going missing is not a crime, but several aspects make it relevant to police. Firstly, missing child reports significantly impact police expenditure (Shalev-Greene and Pakes 2014). Secondly, some children go missing due to serious criminal harm such as child abuse and exploitation, and some witness or engage in crime whilst missing (Newman 1989, Shalev 2011, Simon et al. 2016, Missing People 2019). Finally, some children die, attempt suicide, or experience harm whilst missing. Therefore, preventing harm may reduce missing incidents and vice versa.

To prevent harm, situational crime prevention (SCP) approaches, including supervision, can be effective in reducing child abuse severity and grooming (Leclerc et al. 2015, Wortley et al. 2019, Kamar et al. 2022). However, implementing interventions in all missing child cases is unfeasible, and police data suggests most missing child incidents do not result in harm (NCA 2021). To target effective interventions, context-specific information is required. While there is much research on specific harms suffered while missing (see, for example, Sharp-Jeffs 2016, Simon et al. 2016), or specific missing cases, little is known about the overlap of characteristics and circumstances of harmed missing children. Addressing this gap could help with targeting limited resources to protect the most vulnerable and reduce the impact on police from missing incidents.

This study examined a sample of official reviews of cases where children are known to have been seriously harmed. Child Safeguarding Practice Reviews (CSPRs) document the circumstances of these seriously harmed children. These documents were analysed to identify types of ‘missing’, vulnerability and harm indicators, and associations between those circumstances and guardianship issues. The findings provide insights useful for informing police and partner agency practice for safeguarding missing children in England and provide a basis for expanding research on this issue globally.

Literature review

Missing children

Approximately 65,000 children are reported missing to police each year in England and Wales (NCA 2021), though many incidents go unreported due, it is thought, to parents telling children to leave home, and aversion to police (Rees 2011). Children account for over half of recorded missing incidents, with adolescents comprising the majority of missing children – a pattern replicated elsewhere including the US, Australia and Canada (Huey et al. 2020, Sidebottom et al. 2020, Ferguson and Sidebottom 2022). Multiple UK studies show a recurring pattern of a majority of missing child incidents involving repeatedly missing children (Babuta and Sidebottom 2020, Sidebottom et al. 2020, Galiano López et al. 2023). In Canada, Huey et al. (2020) found more than half of repeat missing incidents were attributed to 16- and 17-year-olds alone. Not only are incidents concentrated amongst adolescents and those missing repeatedly, but incidents have also been found to concentrate in children’s care homes (Hayden and Shalev Greene 2018, Huey et al. 2020). Children in care are overrepresented amongst children reported missing (Hayden and Shalev Greene 2018, Sidebottom et al. 2020, Boulton et al. 2023), and have a higher repeat missing incident rate (NCA, 2021, Sidebottom et al. 2020), though there are more reported incidents involving children missing from home (NCA, 2021). Little is known about the extent to which children missing from care were first missing from home (Sidebottom et al. 2020) – an insight which could help to target police initiatives.

Missing children and harm

Going missing is not necessarily associated with harm. Most missing children are found seemingly ‘safe and well’ (Ferguson and Sidebottom 2022; NCA, 2022). Some missing children are believed to be teenagers ‘pushing the boundaries’ (Hayden and Goodship 2015, p. 451), or children in care¹ experiencing difficulties with their accommodation, or whose carers may be quick to report children as missing to comply with organisational procedures (Newiss 1999, Simon et al. 2016, Sidebottom et al. 2020). A missing incident cannot itself warrant serious interventions, as this demand would outstrip police capacity (Hayden and Goodship 2015). Therefore, identifying characteristics of missing children most at risk of serious harm is important for informing risk assessment and better targeting limited resources towards responses and prevention.

For some children, a missing incident can be an indicator of harm, including child sexual exploitation (CSE), child criminal exploitation (CCE), or abuse that they are running away from (Rees 2011, Hill et al. 2016, Sharp-Jeffs 2016, Simon et al. 2016, Department for Education 2017, Home Office 2017). Child abuse and exploitation are significant issues. In England and Wales, child sexual abuse (CSA) is estimated to have been experienced by 3.1 million adults (ONS 2020), and more than 16,000 potential CSE victims, and more than 10,000 potential CCE victims, have been identified in England (Department for Education 2022). Other harm types associated with missing children include suicide, self-harm, and neglect (Hill et al. 2016, Department for Education 2020; NCA, 2021, Boulton et al. 2023). However, the full picture of the types of, and circumstances surrounding, harm experienced in the lives of missing children is incomplete. Firstly, some studies only report harm experienced whilst missing, excluding harm experienced prior to missing episodes that triggered the incidents (Rees 2011; NCA, 2021, Boulton et al. 2023). Secondly, some statistics, such as those from the NCA (2021), and studies such as Doyle and Barnes (2020), only include recorded harm disclosed, or apparent to, police. Non-police data, for example survey evidence such as the large-scale study by The Children’s Society (Rees 2011), suggests harm experienced whilst missing is more common than the ∼1% reported in NCA (2022) statistics. Children do not always disclose victimisation, or it takes considerable time before they do (Allnock 2015a, 2015b, ONS 2020), and this disclosure is unequal across populations (e.g. boys may be less likely to disclose than girls (Alaggia et al. 2019)). Thirdly, some studies only include missing incidents recorded by police (Boulton et al. 2023). Finally, studies often take a narrow view of harm, examining only a single harm type such as CSE (see Sharp-Jeffs 2016) or specific type of missing such as incidents categorised by police as low or medium risk (see Missing People 2019). This means that a full understanding of the harm experienced by children who go missing is lacking, and current responses to missing children may not address the harm causing their missing incidents, thereby failing to prevent subsequent harm and missing incidents.

In addition to understanding the types of harm experienced by missing children, it is also important (for informing risk assessment and safeguarding measures) to understand the characteristics and contexts of children who go missing and suffer harm. Studies on missing children in Australia and Canada suggest girls are more likely to be victimised than boys (Stevenson and Thomas 2018, Ferguson et al. 2023) but this may misrepresent the true picture as boys are less likely to disclose victimisation (Alaggia et al. 2019). Whilst research on CSE characteristics has focussed predominantly on girls, Cockbain et al. (2017) found that a third of victims in their English study were boys, most of whom had been reported missing and were more likely to have a disability. One of the disabilities included in the study, autism, was overrepresented in their sample compared to the general population and was more prevalent amongst males (ibid.). This led the researchers to call for further research on the link between disability, gender and CSE (ibid.). Other vulnerabilities identified as risk factors, or indicators, for exploitation include a history of abuse or neglect, being in care, unstable/unsafe accommodation, going missing repeatedly, and missing school – specifically absences and school exclusion (Department for Education 2017, Home Office 2017). However, the prevalence of these issues amongst exploited children who go missing, and how they intersect, is unknown.

The location from which children are reported missing (home, care, or school) is an important contextual factor as it may be associated with different harm outcomes and may also highlight different guardianship opportunities. As mentioned above, there is a concentration of missing incidents amongst children’s care homes. Regarding school, children missing from a school setting (including school attendance issues) have been associated with CSE, CCE, and serious violence (Rees 2011, The Children’s Society 2018, HM Government 2018, Missing People 2019) and the UK Government recognises school exclusion as a factor that may indicate, and increase a child’s risk of, sexual and criminal exploitation (Department for Education 2017, Home Office 2017).

Children with special educational needs (SEN), including autism and ADHD², are more likely to be excluded from school than children without SEN, with the highest rates belonging to those without a support plan (HM Government 2022a, McCluskey et al. 2019). Exclusion and SEN have been linked to offending (Department for Education and Ministry of Justice 2022, Arnez and Condry 2021) and emerging evidence suggests a causal link between exclusion and offending (Behavioural Insights Team 2023). However, there is a knowledge gap regarding the interdependency of SEN, exclusion, victimisation, and offending (McCluskey 2019) and researchers call for improved understanding of excluded children’s experiences, including all aspects of their lives, so context-specific interventions can be devised (Arnez and Condry 2021). This gap is wider when including its intersection with missing children. Hence, examining the intersections of these factors in missing children’s lives is important.

Risk assessment and interventions

Police risk assessment

Police risk assessment of missing children may also benefit from a better understanding of the characteristics and circumstances of harmed missing children. The current police risk assessment of missing people in England has been characterised as subjective, inaccurate, and failing to incorporate risk variation across ages, leading to calls for an evidence-based approach (Doyle and Barnes 2020). However, research useful for informing police risk assessment of missing incidents is lacking, with some calling for a better understanding of practitioner constraints in this area (Huey 2019; Ferguson and Sidebottom, 2022).

Situational crime prevention

One framework to prevent criminal harm is situational crime prevention (SCP) (Clarke 2010, Wortley and Smallbone 2010). SCP requires knowledge of specific harm types, and associated circumstances, to inform environmental and managerial changes to be made within, and beyond, the criminal justice system, aimed at reducing crime opportunities, such as through increasing supervision of potential victims (Clarke 1980, Clarke 2017). Understanding the characteristics, vulnerabilities, and harm indicators in missing children who suffer serious harm, can inform prevention initiatives that, through appropriate guardianship, could interrupt a child’s journey to this harm outcome. In fact, a study pre-dating the creation of the SCP approach by Clarke found links between environmental characteristics and children going missing (Clarke 1967, Wortley and Townsley 2017).

A key SCP tenet, rooted in routine activities theory, is the presence of a capable guardian – a formal guardian such as a police officer or an informal guardian such as a member of the public – who can discourage crime by supervising and protecting the victim, thereby increasing the risk to the offender. This reflects the belief underlying the concern with children being missing – a belief supported by evidence from research with offenders that children are unlikely to be able to prevent their abuse themselves (Wortley and Smallbone 2010). SCP approaches shown to discourage harm to children include increasing supervision, and measures leading potential offenders to believe children are supervised (Wortley and Smallbone 2010, Leclerc et al. 2015). Beyond supervision, other victim-focussed SCP approaches proposed to prevent CSA include identifying ‘children with developmental disabilities’ and intervening ‘with lonely, needy, neglected children’ (Kaufman et al. 2010, p. 127 and p. 129). However, implementing SCP may be problematic for issues requiring a co-ordinated multi-agency response – as child protection does (Department for Education 2014) – if practitioners are under-resourced and ill-informed (Clarke 2017). Further, guardians can be unwilling to supervise or intervene (Clarke 1980, Reynald 2010, Schaefer and Mazerolle 2017, Lockitch et al. 2022).

Current interventions

There are two key interventions implemented to address missing child incidents. One approach introduced to improve police risk assessment of missing child incidents is the Philomena Protocol – a police initiative requiring guardians of children in care at risk of going missing to complete a form to provide information to help police to risk assess and find them if they are reported missing (APPG 2019); However, this protocol only applies to children in care. Another approach is conducting independent return home interviews (RHIs). RHIs aim to inform responses to prevent missing child incidents and associated harm, and it is a local authority’s responsibility to conduct an RHI when a child returns from being MFHC (Department for Education 2014). However, challenges to implementing RHIs include local authorities not offering RHIs to every missing child, and offers being rejected (Mitchell et al. 2014, Ofsted 2013, Pona et al. 2019, Boulton et al. 2023). Little is known about RHIs and other responses to seriously harmed missing children. Both RHIs and the Philomena Protocol are aimed at missing children broadly. The current study aims to uncover the contexts specific to missing children who experience serious harm, to inform risk assessment and responses to better target resources to safeguarding the most vulnerable missing children.

Current study

In sum, considerable overlap occurs between missing children and harmed children. Extensive exploratory work examining some elements of this issue exists, but important gaps remain regarding personal characteristics, vulnerabilities, and indicators recognisable in missing children that may signal their path to harm. Understanding these in an intersectional and context-specific manner is vital for developing effective interventions. While interventions exist for some circumstances, there are serious concerns and missed opportunities. Documenting these, in the context of harmed missing children, could provide insights useful for improving safeguarding.

This study aims to overcome the data and measurement limitations described above by examining Child Safeguarding Practice Reviews (CSPRs) using broad definitions of ‘missing’ and ‘harm’ to capture the full range of harms experienced by missing children, including those not disclosed, or apparent, to police. ‘Missing’ comprises children missing from home, care, or school, even if unreported. ‘Harm’ includes confirmed or suspected harm experienced by a child at any time, not solely during a missing incident. This comprehensive approach facilitates a deeper understanding of harm affecting missing children.

Method

Research questions

This study answers the following.

• 1. What characteristics and types of missing, vulnerability, harm and harm indicators are documented in CSPRs, and how do they intersect?

• 2. What guardianship opportunities and issues are documented in CSPRs, and how do these interact with the child’s context?

Data

This study examined Child Safeguarding Practice Reviews (CSPRs) – free-text documents summarising circumstances of children who have died or been seriously harmed, and where abuse or neglect is known or suspected. The child, their family, and safeguarding partners are invited to provide their perspective on the case (Department for Education 2018). CSPRs are not conducted for every case of serious child harm. Reasons to conduct a CSPR include local authorities deeming the cases useful for improving safeguarding practice, recurring themes, multi-agency collaboration issues, concerns about an agency, issues spanning multiple areas, and concerns about safeguarding within institutions (Department for Education 2018). CSPRs are required to be published unless the local authority deems publication inappropriate (Department for Education 2018). Whilst potentially unrepresentative of all cases in which serious harm is known to authorities, published CSPRs enable exploration of the context in which a child has been harmed, from a variety of perspectives, collected by a reviewer (commissioned by the local authority) with no conflict of interest.

The data comprise 39 published CSPRs³ collected mid-2022 - 26 in 2021, and 13 in the first half of 2022 – encompassing 1,474 pages with cases ranging from 14 to 133 pages, with a median of 30. The areas spanned the south, midlands, and north of England, with 11 cases’ authorities unnamed.⁴ CSPRs are publicly available in the NSPCC National Case Review Repository (NSPCC, no date). Permission was acquired from the NSPCC. Cases include fatal and non-fatal outcomes, and feature known or suspected abuse or neglect. Only cases containing substantial details⁵ of children reaching school-age (five to 17 years old) were included.

Analysis

Quantitative content analysis (using a labelling scheme (see supporting material) applied by the first author and an independent annotator) and qualitative thematic analysis of CSPRs were conducted. A broad definition of ‘missing’ was used, with labels including ‘missing from home or care’ (MFHC), ‘repeatedly missing from home or care’ (repeat MFHC) and ‘missing school’ (encompassing any absence from an education setting outside of the home). Additional labels were created for ‘missing school’ subtypes as they were identified in the data.

The labelling scheme also included harm types, exploitation indicators, and vulnerabilities. Harm types that are criminal or require police involvement included those covered by the purpose of CSPRs – abuse, neglect, and death – plus exploitation (a subset of abuse) as ‘missing’ is a potential indicator of this, and each were expanded into subtypes (see supporting material). Since escaping harm can be a reason for a missing incident, harm was labelled whether experienced during, before, or after a missing incident, and (for comparison) for cases that did not mention a child MFHC. CSE and CCE indicators published in UK Government guidance were also included (see supporting material) (Home Office 2017, Department for Education 2017).

Regarding vulnerability, multiple types are used within criminology, and the definition of a particular vulnerability is used in this study – ‘a group or category of people who are more fragile or susceptible to an adverse outcome than others’ – of which subtypes can include situational, innate, social disadvantage and risk (Bui and Deakin 2021, page 2). To inform the list of vulnerability labels, a brief literature search was conducted for risk factors for the harms to children included in this study (Brown et al. 1998, Leclerc et al. 2015, Home Office 2017, Department for Education 2017, University of Manchester 2017, Gotby et al. 2018, Mulder et al. 2018, Stern et al. 2018, Assink et al. 2019, Fedina et al. 2019, Franchino-Olsen 2021, Carballo et al. 2020, Liel et al. 2020, Frederick et al. 2022). These were grouped to condense the labelling scheme (see supporting material).

Labelling guidelines (see supporting material) were created for each label. A stratified (on number of pages quartile) random subsample of 20% (n = 8) cases were selected for labelling by the first author and an independent annotator. The first author’s labels were treated as correct, and both annotators agreed 90% of the time with a Cohen’s Kappa score of 0.76, falling within the ‘substantial agreement’ category as per Landis and Koch (1977). Perfect agreement (Cohen’s Kappa = 1) was achieved for ‘missing’ labels. The first author labelled the remaining cases. Following labelling, frequencies and percentages of cases containing each label were calculated. These were used to provide an overview of the types of missing, harm, vulnerability and exploitation indicators identified, followed by exploratory analysis identifying patterns and associated characteristics relevant to MFHC.

Thematic analysis was conducted to identify recurring themes concerning guardianship opportunities and issues. Whilst analysis includes identifying children missing school due to its connection with MFHC and exploitation, the study’s primary focus is MFHC due to it being a policing issue. In routine activity theory, ‘guardianship’ refers specifically to the presence of a person at the same place and time as a motivated offender and suitable victim, however this study takes a broader view. As explained in Section 11 of the Children Act 2004, police and partner agencies (including local authorities, schools, health bodies, and organisations commissioned by these bodies) are required to give sufficient attention to safeguard children, which includes ‘protecting children from maltreatment’ (Department for Education 2018, pp. 6–7). This obligation exceeds child supervision, therefore ‘guardian’ in this study refers to parents, carers, police, local authorities, health professionals, and school staff. ‘Guardianship’ refers to the act of preventing harm to, and reducing the vulnerability of, children.

Results

Sample overview

The sample contained all harm types (abuse, neglect, exploitation and death), all ‘missing’ types, and all vulnerabilities other than ‘sex work involvement of connections’ (Table 1). 92% of cases mentioned the harmed child missing school. Several ‘missing school’ subtypes were identified, most commonly absence and exclusion. 36% mentioned someone going missing from home or care (MFHC). 31% mentioned the harmed child at the focus of the review going MFHC (69% did not), and of these, most mentioned repeat incidents. Hereafter, referring to MFHC cases includes only those mentioning the harmed child MFHC.

Table 1. Sample statistics – missing, harm, vulnerability.

Label Group	Label (Sub-label)	All Cases (n = 39)		Not MFHC* (n = 27)		MFHC* (n = 12)
		Cases	%	Cases	%	Cases	%
Missing	Missing School	36	92	24	89	12	100
	(Attendance/absence)	23	59	16	59	7	58
	(Exclusion)	12	31	3	11	9	75
	(Home educated)	7	18	6	22	1	8
	(Remote engagement)	6	15	5	19	1	8
	(Off roll / late to enrol)	5	13	3	11	2	17
	(Part-time)	4	10	2	7	2	17
	(Late to school)	4	10	4	15	0	0
	(Sent home/hospital)	3	8	2	7	1	8
	(Missing from school)	1	3	0	0	1	8
	MFHC*	12	31	-	-	-	-
	Repeat MFHC*	10	26	-	-	10	83
Harm	Neglect	32	82	22	81	10	83
	Abuse	31	79	21	78	10	83
	Death	24	62	15	56	9	75
	(Completed)	17	44	11	41	6	50
	(Attempted/near)	11	28	6	22	5	42
	Exploitation	16	41	5	19	11	92
Vulnerability	Child protection services involvement	37	95	25	93	12	100
	Health issues	35	90	25	93	10	83
	Behaviour, police and crime	32	82	20	74	12	100
	Drugs or alcohol	29	74	18	67	11	92
	Domestic abuse	27	69	20	74	7	58
	Economic/accommodation vulnerability	23	59	15	56	8	67
	Social isolation/difficulties	23	59	17	63	6	50
	Stress/stressful life event	22	56	14	52	8	67
	Special Educational Needs and Disability (SEND)	18	46	12	44	6	50
	Parent-child relationship issues	18	46	11	41	7	58
	Prior victimisation/ACE	15	38	8	30	7	58
	Bereavement	16	41	11	41	5	42
	Living with a stepparent	14	36	9	33	5	42
	Lack of supervision	11	28	9	33	2	17
	Sex work involvement/connections	0	0	0	0	0	0

Percentages are rounded. *Harmed child MFHC.

Characteristics, types and intersections of missing, vulnerability, harm and exploitation indicators

All MFHC cases referenced multiple vulnerabilities, multiple harm types and multiple exploitation indicators. All MFHC cases mentioned prior contact with children’s social care either as looked-after children, as part of a child protection plan, or due to a referral to a multi-agency safeguarding hub, and prior contact with police due to missing incidents, domestic abuse at home, offending by a parent or the child, or a combination of these. All but one mentioned drugs in connection with the child or parent (including (mis)use, possession in circumstances suggesting supply, and overdose). 83% featured mental health issues (including depression, anxiety, suicide ideation, and self-harm) of the child or parent. Half the MFHC cases mentioned Special Educational Needs and Disability (SEND), specifically autism or ADHD. Two-thirds featured children experiencing accommodation vulnerability, commonly due to parents/guardians/care placements evicting the child due to their behaviour, or to safeguard the child.

a further seven placement moves […], all of which were in response to the need to manage risks from known, and unnamed, individuals. (Case_25)

Death (including homicide, suicide and attempted/near death), abuse, neglect and exploitation were all common types of harm identified in cases mentioning MFHC. The child died in half the MFHC cases. In three-quarters of MFHC cases the child had died, or death had been attempted/near. Most MFHC cases mentioned multiple harm types known or suspected before the review-triggering harm. In half of MFHC cases, all four harm types were mentioned. Some cases also mentioned high frequencies of harm incidents prior to the review-triggering harm.

 … 130 incidents of self-harming behaviour or suicidal ideation, 13 incidents of [CHILD] being sexually groomed, of which there was evidence in two or three incidents of sexual abuse, and 33 occasions when [CHILD] had gone missing (Case_1, died)

Exploitation was more commonly mentioned in cases mentioning MFHC (92%), than those not mentioning it (19%). Of the cases mentioning exploitation, the majority comprised CSE (44%) or CCE (37%).

In half the MFHC cases, the child disclosed harm prior to the review-triggering harm (see supporting material). Children disclosed to school staff, police, social workers, carers, medical staff, and other children’s parents.

Multiple exploitation indicators were documented in all but one of the cases featuring exploitation of a child in the case (see supporting material). The two most common indicators were injuries, and older or controlling associates, each mentioned in 8 of the 11 MFHC cases mentioning exploitation. All six CCE cases mentioned carrying weapons. Of the seven CSE cases, self-harm was mentioned in four, sexualised behaviour in three, and sexual assault in three.

MFHC cases featured children missing from multiple settings (and often missing school in multiple ways), often multiple times and sometimes incidents were not reported or recorded as missing. In MFHC cases, 58% mentioned being missing from both home and care, and of those mentioning both settings, all were missing from a (grand)parental home first. Care placements from which children went missing, beyond kinship care (including living with grandparents), included foster care, care homes, and semi-independent accommodation.

Of the MFHC cases, the child was missing school in all, with exclusion the most common type of missing school mentioned in 75% of cases. In contrast, of the non-MFHC cases, a minority (11%) mentioned exclusion. This difference is statistically significant, with a Fisher’s exact test score of 0.0002 and p-value less than 0.05. Part-time attendance was more common in MFHC cases (17%) than non-MFHC cases (7%), though this difference is not statistically significant.

Ten in 12 MFHC cases mentioned repeat MFHC incidents. All but two repeat MFHC cases featured exclusion. Of 12 cases featuring exclusion, 75% mentioned the harmed child MFHC, and two-thirds of these involved multiple exclusions. Half of MFHC cases featured SEN – ADHD in four and autism in two – and these children were all excluded, with exploitation known or suspected.

Boys were overrepresented in MFHC cases. 67% of the harmed children in MFHC cases were male and 25% were female. One third of MFHC cases mentioned unreported/unrecorded missing incidents, and of these four, three featured boys. Incidents were unreported by both parents and professionals. In one case mentioning CCE, the boy was often recorded as ‘wanted’ by police rather than missing, and some care staff did not report him missing (Case_25). All CCE cases featured boys and all but one of the CSE cases featured girls. Table 2 summarises harm types, MFHC, exclusion, SEN, age, and sex for the non-MFHC cases mentioning exploitation and all MFHC cases.

Table 2. Harm types, MFHC type, exclusion, SEN type, and demographics in non-MFHC cases mentioning exploitation and all MFHC cases.

Case	Harm Type				Missing Type		Vulnerability	Demographics
	Abuse	Neglect	Death	Exploitation	MFHC	Exclusion	SEN	Age 11+?	Sex
	MFHC Cases
1	Yes	Yes	Died & att/nr	CSE	Repeat			Yes	Female
3	Yes	Yes	Died	CSE	Repeat	Multiple	Autism	Yes	Female
37	Yes	Yes		CSE	Repeat			Yes	Female
31	Yes	Yes	Att/nr	CSE/CCE	Repeat	Multiple	Autism	Yes	Male
39		Yes	Died	CCE	Repeat	Multiple	ADHD	Yes	Male
23	Yes	Yes	Died	CCE	Repeat	Yes	ADHD	Yes	Male
28	Yes	Yes	Att/nr	CCE	Yes	Multiple	ADHD, learning difficulties, conduct disorder	Yes	Male
16			Died	CCE	Repeat	Multiple		Yes	Male
25	Yes	Yes		CCE	Repeat	Yes		Yes	Male
24	Yes	Yes	Att/nr	General	Yes			Yes	Male
38	Yes	Yes		General	Repeat	Multiple	ADHD	Yes	Multiple children
27	Yes		Died & att/nr		Repeat	Yes		Yes	Male
	Exploitation Cases (Not MFHC)
30	Yes	Yes		CSE		Multiple	Autism suspected	No	Female
8	Yes	Yes		CSE				No	Male
33	Yes	Yes		Mother CSE				Yes	Female
13	Yes	Yes	Died	Radicalisation		Multiple		Yes*	Male*
12	Yes	Yes		General				No	Female

*Harmed child was younger than 11, but sibling who was radicalised was 11 and male.

Guardianship opportunities, issues, and their interaction with the child's circumstances

The findings highlight extensive knowledge available to guardians about the vulnerabilities, harms and harm indicators associated with missing children in the sample, presenting multiple opportunities to intervene to prevent harm and further missing incidents. However, information was frequently underutilised, owing to information not being shared, a single-incident focus, or insufficient time to read information.

 … professionals for child protection case conferences locally are required to read the reports provided to the conference in the 30 minutes before the conference starts […] caused by guidance not always being followed regarding submitting reports days in advance enabling all to prepare. This means that it is difficult for all professionals to fully understand complex circumstances (Case_9)

 … incident was closed by the police without obtaining more information which would have revealed the historical and current concerns […]. Had a missing report been submitted to Children’s Services […] this would have prompted the offer of a Return Home Interview (Case_24)

Information usage and sharing issues also affected RHIs which were not always conducted, sometimes due to declined offers or police not sharing missing incidents with local authorities. Ineffective use of information from interviews was also highlighted, such as only focussing on single incidents rather than considering other information about children, and failing to use information to develop safeguarding plans for a child or to analyse patterns to inform strategy.

 … independent return from missing interviews are not being consistently completed, that incidents tended to be processed in isolation rather than assessed cumulatively, are not being used effectively enough to help develop risk strategies for individual children or being aggregated and collated to help identify potentially risk hotspots … (Case_1, died)

Underutilising information sometimes led to ineffective safeguarding, with children’s behaviour perceived as the problem rather than any harm the behaviour may indicate – illustrated in the excerpts below.

[CHILD] went missing on four occasions after she was ‘closed’ to [SERVICE] […] police assessed two of these missing episodes as medium risk, and as [SERVICE] only dealt with high risk cases, these were not referred back into the [SERVICE]. The risk to [CHILD] was clearly viewed as a one off episode rather than an early indication of sexual exploitation. (Case_3, died)

 … this should have included gathering information about [CHILD]’s childhood and family experiences, but this information is not available on records […]. Consequently, [CHILD] was perceived to be the problem and services were provided to address his behaviours. (Case_25)

Reviewers highlighted the importance of information sharing and usage due to the ‘piecemeal manner in which young people will tend to disclose information’ (Case_1) and how an incomplete understanding of the child’s circumstances during risk assessment resulted in plans that ‘did not address all the contexts in which [CHILD] was likely to experience harm’ (Case_27, died).

Insufficient intervention following harm disclosure was a recurrent theme. Issues included disbelief, delayed or no safeguarding action, and insufficient action such as relying solely on the child’s disclosure and cooperation with investigation, highlighting the importance of using all available information about a child’s circumstances.

 … there was a strong focus on victim disclosure so that despite there being overwhelming evidence that [CHILD] was a victim of CSE, the over reliance on a specific allegation meant that professionals did not make a clear determination that [CHILD] was a victim of CSE (Case_37)

Although there had been a plan to issue a section 2 notice after the first disclosure no such notice was issued (Case_1, died)

Guardians often viewed themselves as unable to supervise the child due to the risk the child posed to themselves, to other children, or to the guardian. This was commonly due to the child’s behaviour, which was often an exploitation indicator such as self-harm, sexualised behaviour, carrying weapons, or associating with dangerous people.

[CHILD] stayed at [CARE_HOME] for around 7 weeks before CSC was given 7days notice that the placement was to be terminated. […] because of her sexualised behaviour, posed a risk to other residents (Case_3, died)

Despite daily searches no suitable accommodation could be found. […] it was recognised that any staff ‘supervising’ [CHILD] would be at significant risk if they tried to curtail his liberty. By this time information had come to light that [CHILD] was using firearms (Case_25)

This issue was also associated with school exclusion. Within MFHC cases mentioning exclusion, most mentioned behaviour as the reason, including behaviour classed as ‘challenging’, threats, assaults, and ‘sexualised behaviour’. In one case, the exclusion was due to a CSE victim writing about an abusive sexual relationship:

As part of an English assignment […] [CHILD] produced a piece of writing […] in the first person, the text describes an abusive sexual relationship […]. There is also a reference to ‘my sexual abusive past’ […] [CHILD] was excluded for three days for submitting this piece of work (Case_3, died)

In one case, exclusions were linked with unsupported SEN:

 … his exclusions appear to have happened during times where he was not receiving SEN support for Behavioural, Emotional and Social Difficulty (Case_39, died)

Insufficient training (e.g. for self-harm, exploitation, and complex cases) was also a common theme and intersected with harm, and exploitation indicators, leading guardians to deem children too risky to supervise:

Training expectations […] are that clinical and nonclinical staff who come into contact with people who self-harm have a sufficient understanding […] several of the carers reported that they had not had that input or felt insufficiently prepared. (Case_1, died)

Guardians perceiving harmed missing children as too risky often resulted in the child's eviction from their home or care placement by their guardian, or exclusion from school. Regularly moving home or care placement (accommodation instability) and school exclusion are risk factors for exploitation. Vulnerability was potentially increased further when care placements children were moved to (such as semi-independent accommodation) had less supervision and support. This was sometimes done despite advice or attempts to move the child to a placement with increased supervision and support.

 … though there was discussion about identifying a bespoke care placement for [CHILD] and the [MENTAL HEALTH] service repeated the call made by community CAMHS […] that [CHILD] should be in a care placement with in-house therapy, she returned to the children’s home and was transferred to the semi-independent unit … (Case_1, died)

 … the main challenge […] was finding a specialist resource that could and would hold him and manage his challenging and violent behaviours. Moving [CHILD] to semi-independent placements brought […] limited if any oversight of his […] whereabouts. [CHILD] went missing from these placements on a regular basis (Case_25)

Discussion

To explore how the characteristics and types of missing, vulnerability, harm indicators and guardianship issues feature and intersect in cases of serious harm, this study analysed 18 months of CSPRs published by English authorities. Although most cases did not mention children MFHC, the 31% that did exceeds the less than 1% of reported missing schoolchildren in England, indicating disproportionate representation. The findings show how seriously harmed children MFHC were commonly missing repeatedly, commonly missing from home before missing from care, missing school in numerous ways, and most were excluded from school. All mentioned multiple vulnerabilities, prior contact with police and social care, and the majority mentioned drugs, mental health issues, and accommodation vulnerability most commonly due to frequent care placement moves. SEN was overrepresented compared to the general population ⁶ with half the MFHC cases mentioning ADHD or autism. All mentioned multiple harm types with most mentioning abuse and neglect, and all but one mentioned exploitation.

These circumstances presented multiple opportunities for safeguarding (disclosure, exploitation indicators, knowledge or concerns of harm and vulnerabilities, and multiple missing incidents), especially within schools due to the high prevalence of exclusion and the numerous ways in which children were missing school. However, the combination of these factors with guardianship limitations (with information usage, information sharing, and training) often resulted in responses increasing vulnerability through exclusion, eviction, and reducing supervision and support due to the child being seen as too risky to others but, paradoxically, sometimes treated as not risky enough to meet the threshold for support. This situation may more greatly impact certain groups of children, specifically boys and CCE victims (due to being less likely to disclose victimisation and who may be more likely to be seen as risky offenders than vulnerable victims), and children with SEN (due to an increased need for support and overrepresentation amongst children excluded from school).

Implications

Previous research (Sidebottom et al. 2020) shows that many missing child incidents are generated by a repeatedly missing minority. The current study found most MFHC cases of seriously harmed children involved repeat missing incidents. Directing policing resources towards the minority of repeatedly missing children may provide police with opportunities to not only reduce a substantial proportion of missing incidents but also safeguard those most seriously harmed. To do this, police need to know information about missing incidents, vulnerabilities, harms experienced, and harm indicators. This kind of information was known to guardians for the missing children in this study. If this finding generalises, it reveals an opportunity for police to target these seriously harmed and repeatedly missing children. However, the findings also showed that the existence of this information is insufficient. Information needs to be retrieved and used effectively, requiring trained practitioners to know the information needed, and sufficient time to use it, which often was not the case in this study’s sample. The impact of these issues may be exacerbated for children such as those in this study due to the numerous vulnerabilities and harm types, as practitioners are likely to require additional training and time to conduct an accurate risk assessment and respond effectively.

Issues with police risk assessment found in this study mirror those raised previously (Hayden and Goodship 2015). Insights from RHIs could be important for police to accurately risk assess subsequent missing incidents. The findings highlighted issues with inconsistent implementation of RHIs, insufficient information-sharing, and ineffective usage of insights gained from RHIs, adding support to findings from previous research (Ofsted 2013, Chetwynd and Pona 2017, Pona et al. 2019, Boulton et al. 2023). This study showed how these issues have led to ineffective safeguarding of seriously harmed children, due to some missing incidents not being treated as high risk despite the same children being seen as too risky to supervise due to exploitation indicators. Whilst addressing training and gaps in specialist care provision is needed, better police risk assessment of missing child incidents through proactive information gathering and a child-centred (versus single incident centred) approach, and earlier intervention, may reduce the number of children whose harm escalates to the point of being perceived as too risky to supervise.

To improve risk assessment in the context of insufficient data sharing, police should assume that not all information has been previously shared, and proactively request information from partner agencies upon receiving missing child reports. Given issues with insufficient training, it may be useful for guidance and forms associated with missing child risk assessment to contain specific prompts about the types and sources of information useful for identifying harm.

Schools could assist with police risk assessment, by completing the Philomena Protocol form (currently only implemented for children in care) for children they are concerned about such as those at risk of exclusion. This is particularly beneficial for children in care whose carers know them less well than school staff due to frequent changes in care settings, as was commonly the case for missing children in this sample. This may also benefit children missing from abusive or neglectful homes whose carers may be unwilling or unable to provide the information necessary for an accurate risk assessment. Additionally, it would allow for better risk assessment of children who initially go missing from home before they go missing from care. Little is known about whether children missing from care went missing from home first and this insight would help to clarify the relationship between care and going missing (Sidebottom et al. 2020). This study begins to address that gap as it found children missing from both home and care accounted for more than half of the MFHC cases in the sample and all went missing from home first.

Boys MFHC were overrepresented in this sample, aligning with the proportion of boys in Serious Incident Notifications (HM Government 2022b) but contrasting with studies in Australia and Canada finding victimisation more prevalent for missing girls (Stevenson and Thomas 2018, Ferguson et al. 2023). Each boy MFHC in this sample also experienced multiple types of harm. Multiple harm types experienced by girls MFHC were highlighted by Ferguson et al. (2023) but not boys. Possible reasons for this disparity include (but are not limited to) cross-national differences, boys being less likely to disclose victimisation, or more missing boy incidents being unrecorded, but further research is needed. Since all cases mentioning CCE in this sample involved boys, fear of punishment of crimes committed whilst exploited may be a factor in fewer disclosures of victimisation by boys. Further, being recorded as ‘wanted’ rather than ‘missing’ would exclude their victimisation from research on missing boys.

Many of the common types of vulnerability co-occurring with MFHC cases in the current study – police involvement, child protection services involvement, prior victimisation, accommodation instability, drugs, mental health issues, SEND – resonate with previous research (Biehal et al. 2003, Crosland and Dunlap 2015, Ferguson et al. 2023). The prominence of vulnerabilities in the current study that were either driven by decisions made by formal guardians (e.g. care home eviction and school exclusion), or ineffectively supported by formal guardians (e.g. lack of support for SEN), highlights potential opportunities for reducing risk and missing incidents. This is particularly pertinent for children with ADHD or autism who were overrepresented in MFHC cases in the current study, all of whom were excluded and suspected or known to have been exploited. This supports previous literature showing children with SEN are excluded from school at a higher rate and autistic children are overrepresented in CSE cases (Cockbain et al. 2017, HM Government 2022a).

Finally, this study’s findings support Clarke’s (2017) contention that ill-equipped multi-agency responses may hinder SCP, by highlighting how an SCP approach is obstructed by resource and training deficiencies, exacerbating the vulnerability of highly vulnerable, seriously harmed missing children. The findings also support studies highlighting guardians’ unwillingness or incapability to supervise in certain circumstances (Reynald 2010, Lockitch et al. 2022). Given repeat missing incidents are an exploitation indicator, guardians’ unwillingness to supervise children exhibiting behaviour indicative of exploitation (due to insufficient training or resources to manage the risk the child poses to themselves, other children, or the guardian) may not only be a barrier to preventing exploitation and missing incidents but may be a facilitator of those issues. Future research should examine barriers and facilitators to safeguarding children exhibiting behaviour potentially harmful to themselves or others, and how this might impact the likelihood of the child missing school or going MFHC.

Limitations and future research

This study has several limitations. First, it excludes unpublished CSPRs and may be unrepresentative of all English CSPRs. Second, CSPRs are not restricted to reviewing a single child and can contain references to others who have been missing. Therefore, findings are relevant to the case, not a specific child, unless otherwise stated. Third, missing incidents and other events may be undocumented. Further, the data cannot answer questions about unharmed missing children, those whose harm is undetected by authorities, or cases outside a CSPR’s scope. Finally, the sample size is small. Thus, the study abstains from claiming generalisability.

Despite limitations, this study presents novel insights into the characteristics and intersections of missing, vulnerability, harm indicators and guardianship issues associated with seriously harmed missing children, revealing several avenues for future research. Researchers should explore whether victimisation of missing boys is more prevalent than currently thought. Second, the extent to which the framework of common vulnerabilities, harms, indicators of exploitation and types of ‘missing’ found in this study represents the wider population of seriously harmed missing children should be examined, to identify if this is an indicator useful for risk assessment and prevention. Third, including the school setting within research on missing children (ideally with a comparison group of children not MFHC) may improve knowledge of how MFHC, exclusion and related factors such as autism and ADHD, interrelate with victimisation. Additionally, future research should study police risk assessment of missing children and how support services use it. Solutions to practitioners having insufficient time to use all available information about missing children to inform accurate risk assessments also requires exploration.

The study emphasises the need for a more comprehensive, proactive and child-centred approach to understanding and protecting missing children from harm, acknowledging guardianship barriers, including insufficient training and time to fully understand a child’s circumstances, and unwillingness to supervise children showing exploitation indicators. Schools could help to improve police risk assessment of, and identify risks and harm to, children at risk of going missing. Boys, CCE victims, and children with SEN may benefit most.

Acknowledgements

The authors would like to thank David Buil-Gil, Laura Bui, Diana Bociga, and Ezra Lampesberger for their valuable feedback on earlier drafts of this paper.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The dataset used in this study is publicly available via the NSPCC National Case Review Repository.

Additional information

Funding

This work was funded by the School of Social Sciences, University of Manchester.

Notes

1 A child in England is ‘in care’ when removed from their home by a local authority, most commonly due to abuse or neglect, and formally placed under the care of another guardian e.g., a relative (kinship carer), a foster carer, or the local authority accommodated in a residential children’s care home or in semi-independent accommodation.

2 Attention Deficit Hyperactivity Disorder.

3 See supporting material.

4 NSPCC has clarified that unnamed authorities are English.

5 Documents containing only executive summaries, reviews only containing a summary of recommendations, reviews focussing on offenders, and thematic reviews not containing substantial details of the circumstances of individual children were excluded. One CSPR contained 2 unrelated cases, with one child of school age, and one child who was not. In this case, only details provided for the school-aged child were reviewed.

6 In 2021/22 in England, less than 13% of schoolchildren received SEN support (HM Government 2022c).