We need to talk about CSE Toolkits

Is the child you are working with low risk of CSE? High risk? Medium risk?

Is the child you are working with low risk of child sexual exploitation? High risk? Medium risk?

Would they be the same ‘risk level’ in another authority?

Does their recorded risk level match your professional judgement?

What does ‘risk level’ really mean, anyway?

In August 2016, the Home Office commissioned the Early Intervention Foundation and Coventry University to undertake a rapid evidence assessment entitled: ‘Child Sexual Abuse and Exploitation: Understanding Risk and Vulnerability’. The results of that research were released earlier this month.

The executive summary included the assertion that risk indicators in child sexual exploitation, vulnerabilities, protective factors and interventions were based on little to no evidence base. The second point was that risk indicator toolkits in CSE are plentiful, but have no evidence base, have not been evaluated and have not been developed using rigorous methodology.

Anyone who has been on one of our JustWhistle training courses or seen us speak at conferences in the last few years will know that this is something we have been arguing for a long period of time. We have been calling for the rationalisation, testing and validation of these tools for a number of reasons:

1.       The toolkits were developed without any evidence base but are being used as a psychometric measure of risk of CSE

2.       Indicators of CSE are themselves conforming to a narrow stereotype, leading to professionals looking at particular groups of children and completely missing others

3.       Risk indicator toolkits are too prescriptive,  and can lead to decisions and actions that bypass the professional judgement of the workers or parents/carers who know the child well

4.       The language used on risk indicator toolkits is not only incorrect, but is actively mixing up risk with victimhood and therefore failing to protect children

Therefore, we were delighted to see that the results of this new report from the EIF and Coventry University support our longstanding argument and will draw much needed attention to the issue of using untested, unproven prescriptive tools to direct the safeguarding response to children who are being sexually exploited.

This article will expand on the four points made above using evidence from the rapid evidence assessment.

1.       The toolkits were developed without any evidence base but are being used as a diagnostic measure of risk of CSE

CSE toolkits started to be developed in around 2006, but charities and organisations started to write about CSE indicators and CSE protective factors as early as 2001. Over a period of time, organisations working in the field of child sexual exploitation started to release sets of CSE indicators designed to support professionals in identifying children who could be at risk of CSE. Whilst many practitioners in the field at the time will agree that they gave much needed guidance and parameters to the difficult job of identifying and protecting children who were at risk of being exploited, it is reasonable to say that they were not rigorously developed, tested or validated (Brown et al., 2016).

As the message of ‘any child, anywhere’ spread across the UK, so did the usage of CSE indicator lists. Quickly, hundreds of authorities and organisations were utilising a completely untested set of ‘risk indicators’ to use as a checklist of symptoms of child sexual exploitation. As the systems have become more sophisticated and as teams have started to work together to tackle child sexual exploitation, risk indicator lists that were once used to loosely guide professionals turned into mandatory, rigid, diagnostic scoring systems within which, indicators were labelled ‘low’, ‘medium’ or ‘high’ risk of CSE occurring without any study taking place to look at whether that labelling was accurate. Not only were indicators resigned to this labelling, but children too, were beginning to be labelled as ‘low’, ‘medium’ or ‘high’ risk of CSE depending on how many of the relevant indicators were ticked off on the checklist, teamed with the vulnerabilities and the protective factors, that were later added to CSE toolkits. An important example of why the lack of testing is such a problem is the point made by Brown et al. (2016, p.16) where it is stated that ‘most of the studies identified in the review did not compare victims with non-victim groups, or use any other methodological designs that allow us to identify variables that indicate increased risk’.

Essentially, this point is being made because when previous authors have claimed to have interviewed, conducted surveys or collected data from casefiles about indicators of CSE, they have not then conducted the same comparative set of studies from a group of children who had never experienced CSE or CSA. This second set of children would act as the control group, meaning that it would then be easier for the researchers to see whether the indicators were unique or more prevalent in the children who had been abused or whether some indicators were equally common in both groups of children and therefore could not be reliably called an ‘indicator of CSE’. Our experience of teaching thousands of professionals across the country has taught us that professionals deal with this issue frequently. They work with children who are ‘fulfilling’ particular risk indicators and then it turns out to be another safeguarding or mental health issue – or on the reverse – they work with children who do not fulfil enough risk indicators and it turns out that they were being abused all along. Because the indicators have not been tested, it is not clear if they are accurate. Post empirical testing, some ‘classic’ indicators may remain, others may not and new indicators may be suggested.

The reality is that we continue to use diagnostic toolkits under the presumption that there is proven causal relationship between the risk indicators and the risk of CSE. Rigorous study is the only way to validate and test whether risk indicator toolkits are doing what they say on the tin and we welcome this recommendation from the authors.

2.       Indicators of CSE are themselves conforming to a narrow stereotype, leading to professionals looking at particular groups of children and completely missing others

One argument that seems to be getting more and more airtime is the accurate observation that CSE toolkits are female-centric and it is therefore nearly impossible for a boy to be categorised as high as a girl using the same toolkit because some of the indicators are very clearly female-related (‘highly sexualised dress’, ‘repeat terminations’, ‘miscarriage/pregnancy’, ‘suddenly wearing make-up’) and other indicators are affected by a broader gender stereotype in which boys would not be seen as at as much risk as a girl simply due to the embedded assumption that they are stronger, more ‘streetwise’ and more likely to be in gangs and crime. This is just one small part of a wider issue with CSE toolkits.

If the toolkits are examined closely, they represent the exact stereotype of a child that JustWhistle have been working to deconstruct for years. They list children with chaotic families, poor parenting, drug use, alcohol use, and poverty, performing poorly at school, going missing, wearing sexualised clothes, getting in stranger’s cars and becoming aggressive. Yet despite these lists, professionals will consistently report that the children they work with do not conform to that list. The children they are working with are diverse, they are intelligent, they are from all different socioeconomic backgrounds, their parents are totally engaged in the safeguarding process, they are performing well at school and they didn’t show any obvious changes in behaviour at all. They were incredibly difficult to identify and can be equally as difficult to protect. This disconnect between the risk indicators and reality means that children who conform to all of the stereotypical indicators of a vulnerable child will be more likely to receive a response than a child who is also being sexually exploited but doesn’t conform to the ‘norm’.

This argument is broadened by Brown et al. (2016, p.7) who as part of their recommendations, argue that giving professional stereotypical symptoms of CSE creates on over-reliance on those indicators and should be discouraged. This can also be taken to support the earlier example of female-centric toolkits, which are so stereotypical and so narrowly focussed on the female victim that boys and young men are not being referred as quickly, are not being referred in the same way and are often known by criminal justice agencies before they are ever identified as a victim of CSE (Cockbain, 2015).

3.       Risk indicator toolkits are too prescriptive, too dehumanising and bypass professional judgement of the workers or parents who know the child best

There are two main issues raised in the piece of research from the EIF and Coventry University that are related to this particular point. The first is the prescriptive nature of the toolkits and how they can lead to conclusions being drawn about children, and how this may then influence how they are responded to, what’s in their support plan and their ‘risk level’.

Of the ten individual toolkits reviewed as part of the review, none had ever been tested and evaluated for validity and reliability. Critical decisions are being made about children all over the UK using frameworks that have never been tested for accuracy, long-term outcomes, predictive power or reliability across children or for the same child across their lifespan.

It is imperative that when a system, psychometric measure, behavioural measure, self-report evaluation form or a risk indicator checklist such as this one is developed, that it is tested on large diverse samples of children to ensure that it works. Whilst we do not have this evidence, we should not be using these toolkits – or at the very least, using them with extreme caution. We should be questioning guidance that outlines ‘if a child is medium risk then this must happen and we must do these things and this child must undertake these pieces of work…’ because there is nothing to support that it will work or that the conclusion drawn from the toolkit it the correct one.

The second criticism raised in the piece of research which is relevant to this point is ‘the discouraging or elimination of professional judgement and decision making with an over-emphasis on scoring’. Again, this is a concern JustWhistle has been raising for a number of years now and those of you that have been on our training or heard us speak will know that we frequently advise professionals to escalate the concern and to argue that professional judgement and human knowledge of the child outweighs the result of a checklist.

Indeed, there are some toolkits in the UK that do not allow the professional to have any comments entered on to the form and there are some toolkits that are slowly moving towards a model where the indicators are still prioritised, but the professional is able to argue their point, add their own context and highlight the significance of a particular indicator if they think it will improve the response and outcome to a child. This finding should be treated as an urgent development in the field and should be used to give power back to the parents and the professionals that know the child best. Whether that is when a parent is arguing that their child is at much higher risk than the toolkit assessment is indicating or whether that is a professional who has been working with a child for over a year who is trying to show that their risk level has decreased and certain measures need to be reduced.

Context and professional judgement is key. We cannot and should not attempt to quantify the experience of abuse within a checklist whilst losing the voice of experienced parents and qualified professionals.

4.       The language used on risk indicator toolkits is not only incorrect, but is actively mixing up risk with victimhood and therefore leading to the failure to protect children

Out of all of the issues that are raised in this piece of research, this finding could be one of the most important. It is something that JustWhistle have been arguing for years. The language of ‘risk’ in CSE is so far removed from the English definition of the word that the meaning and purpose of the toolkits has been lost. Risk is defined as ‘a situation involving the possible exposure to danger’ and ‘the possibility that something unwelcome or unpleasant will happen’ (Merriam-Webster, 2016). Both of these definitions also contain synonyms of risk which include ‘possibility’, ‘likelihood’, ‘chance’, ‘probability’ and other words that would be used to describe something that has not yet occurred. In the field of safeguarding, risk is used to define something that may happen. Risk assessments are a formula to consider what may happen or what may pose a threat to a client, to a worker or to other people so that it can be planned for and if possible, prevented.

In CSE, ‘risk’ is used in three categories – low, medium and high. Low risk tends to include situations and behavioural changes that would indicate that something was already affecting the child: regularly going missing, meeting with unknown adults, self-harming and eating disorders. These indicators may not be related to CSE but they are a sure sign that the child is suffering from something. However, when we move to look at medium and high risk indicators on most tools, the ‘risk indicators’ are quite clearly describing a child that has gone far beyond ‘risk’. They are describing a child that is already a victim of CSE: Having a much older partner who is known to be a risk to children, associating with known CSE perpetrators, associating with known CSE victims, being found at CSE hotspots, making an unsubstantiated disclosure of sexual assault – and those are just the ‘medium risk’ indicators.

The ‘high risk’ indicators that are supposed to describe a child that is at high risk of CSE occurring but it has not yet happened include: child under 13 coerced into sex, child trafficked for sex, peer on peer abuse, being taken to licensed premises for sex with adults and being sold. Brown et al. (2016, p.7) were absolutely right in saying that some ‘indicators of CSE are actual signs of abuse or exploitation already occurring’. Therefore, a child can be actively being sexually exploited and trafficked and still being categorised as ‘high risk’. Again, Brown et al. (2016, p.6) make the argument that the threshold for being an actual ‘victim’ is far too high. So it begs the question, if a child can be abused and sold and they are still high risk, when will they be considered a victim?


This article was written to highlight the real world impact of the findings of this new publication. With an entire system, whole policies and procedures and even staff teams built around the assumptions and issues discussed in this article, it is imperative that this research is shared and discussed at every level. Whether you develop policies in CSE for your LSCB or whether you are a professional working frontline with children who have experience CSE – these findings impact all levels of our work and need to be considered.

We constantly review research and adapt our approach in training so participants have the most up to date information available to them. This is so that they understand recent trends, key questions in the field and the challenges of day-to-day practice.

The validity of risk assessment frameworks and indicator checklists has been an issue that has challenged us and other practitioners for a long period. We understand why they were initially developed, however we have consistently challenged their validity and the practice we observe and have reported to us on their usage and their weaknesses. We commend the Home Office on commissioning this research and the EIF in partnership with Coventry University for the findings; findings which speak for themselves. The challenge is now two-fold. Commissioners of research and service delivery in CSE must extend this work by offering opportunities to find improved ways of identifying, evaluating and assessing risk in CSE. Meanwhile, frontline workers and management have to accept the findings and review how and why these untested frameworks are being employed with children without an evidence base.

This article was written by Jessica Eaton (Training Manager) and Nicola Dalby (Business Manager) for Safe and Sound Group.



Sign up for our newsletter to receive updates, as well as educational and awareness raising articles.

Sign up

Support us

You can make a one-off donation or set up a monthly direct debit to us at any time. Your money will go directly towards our one-to-one support service for young people affected by child sexual exploitation.


Live tweets