Experiencing abuse, neglect, bullying, or domestic violence in childhood increases the likelihood of having poor functioning in young adulthood, but this is not the case for everyone. Being able to accurately predict which individuals are at high risk for poor outcomes following such negative childhood experiences could support professionals to effectively target interventions. Is it possible to make accurate personalised predictions? And if so, how acceptable and feasible is it to use these in health and social care practice?
The context
It is a sad reality that almost one-in-five British children experience abuse, neglect, bullying or domestic violence (Radford et al., 2013). Not only are these experiences incredibly distressing at the time, but they can continue to have impacts into adulthood. On average, these children have lower educational attainment, they are more likely to have criminal convictions, to be a teenage parent, and to feel lonely and socially isolated compared to their peers who did not have these negative experiences (Currie & Widom, 2010; Herrenkohl et al., 1998; Jaffee et al., 2018; Malvaso et al., 2018; Matthews et al., 2017).
Fortunately, this is not the case for everyone. This ‘average pattern’ disguises the fact that there is huge variation in how individuals respond to negative childhood experiences, with some people doing much better than expected despite what they have been through (Cicchetti, 2013). However, it is unclear if it is possible to predict which children will be resilient following negative experiences and which children will have poor outcomes.
Answering this question could help professionals working with these children – like social workers, health care practitioners, and teachers – to identify those who are most at risk of having poor outcomes so they can ensure that these vulnerable children receive the support they need to flourish.
This approach is already common practice in medicine. GPs use a tool called a ‘risk calculator’ to find out their patients’ risk of developing a particular physical illness, such as breast cancer or heart disease. They enter specific information about the patient – for heart disease predictions this includes details of their cholesterol level and whether they smoke – into an algorithm which then calculates that person’s level of risk. This risk score can help the GP decide whether an intervention, like medication or a lifestyle change, is needed to try and prevent the illness developing.
A new direction
Taking inspiration from this, we attempted to develop a risk calculator to be used with young people who were exposed to abuse, neglect, bullying, or domestic violence during their childhood (Latham et al., 2019). This uses information about them, their family, and their community to calculate their individual risk of having poor outcomes as a young adult.
To create this risk calculator, we used data from a study of over 2000 children who were born in England and Wales in 1994-1995 and have been followed since birth until 18 years of age. During this time the research team collected lots of information about them including information on their negative childhood experiences, their personality, mental health symptoms, family environment, and community environment.
When participants were 18 years old, they were interviewed about their psychosocial wellbeing, answering questions about how satisfied they were with life, loneliness, social isolation, experiences of being victimised during adolescence, and their quality of sleep. They were also asked about their economic situation – including what educational qualifications they have, whether they were currently in education, training or employment, and whether they were a parent. Participants also gave permission for the researchers to access official police records of any criminal convictions or cautions that they had.
Using this data, we found that a different combination of childhood information is needed to predict poor psychosocial and poor economic outcomes:
Having poor psychosocial outcomes was best predicted by a combination of:
• Individual factors (being female, having conduct disorder, anxiety, and self-harm or suicidal thoughts in childhood)
• Family factors (a family history of mental health problems, not having a warm relationship with their mother, and not having an adult to turn to for support)
• Community factors (living in a neighbourhood with high levels of crime and no sense of community, and having poor quality friendships)
Having poor economic outcomes was best predicted by a combination of:
• Individual factors (being male, having lower intelligence, particular personality traits, and the presence of attention deficit and hyperactivity disorder symptoms in childhood)
• Family factors (coming from a poorer family, not having a warm relationship with their mother or siblings)
Next, we tested whether the risk calculator’s predictions were accurate. Because the children in the study had been followed until they were 18, we were able to compare what we know about their psychosocial and economic outcomes at this age with the predictions made in childhood by our risk calculator, to see how well they match.
We assessed this using the C-statistic, a measure of how well the risk calculator can distinguish children who do develop poor psychosocial or economic outcomes from those who do not. This statistic can range from 0.5 (which represents chance-level discrimination) to 1 (which represents perfect discrimination). The C-statistic for predicting economic outcomes was 0.80, this showed us that the risk calculation was very good, it performed much better than would be expected by chance! The risk calculation for psychosocial outcomes was not quite as accurate (C-statistic = 0.65), but this also still performed better than chance.
The next important step will be to test how accurate the risk calculator’s predictions are when it is used with a different sample of young people who have had negative childhood experiences, such as those already under the care of social services.
Considerations for implementation
The future implementation of a risk calculator may not be straight-forward so it is vital to understand what the key issues are from those who may use it. To investigate this, we interviewed 13 UK professionals from health and social care who work with children exposed to abuse, neglect, domestic violence, or bullying. We also spoke to a group of 6 young people (aged 18-21 years old) who had lived experience of these negative events in their childhood.
There was a lot of similarity between the views expressed by the professionals and the young people (Latham et al., 2020).
They believed that an accurate risk calculator had potential benefits, including:
• Encouraging a shift towards early intervention and prevention of poor outcomes among those exposed to abuse, neglect, domestic violence, or bullying
• Helping professionals to evidence the need for, and secure, intervention for a child
• Providing a common focus and language for the various professionals working with a child
• Helping professionals to better allocate limited resources to those who most need it
However, they also highlighted some important considerations and challenges, such as:
• Concern about whether risk calculations would be accurate
• The potential for a high-risk score to evoke negative feelings in the child (e.g., hopelessness) and their caregiver (e.g., guilt, blame)
• Whether professionals should use the risk calculator with the child and their caregiver and share the risk score, or use it without their involvement simply to inform professionals’ assessments and work with the child
• The necessity for resources/services to be available to effectively support children who are identified as being at high risk of poor outcomes. Without this, professionals were unwilling to use the risk calculator
• The need for practitioners to be trained to use the risk calculator and interpret the risk score
Conclusion
We hope that this work shows the potential for a risk calculator to help professionals working with children who have had negative childhood experiences to identify those who are at highest risk of poor psychosocial and economic outcomes so they can ensure these children receive the support they need. It is, however, important to remember that children who receive a low risk score from our calculator may still be at risk for developing other poor outcomes, for example mental health problems like anxiety, depression, and substance misuse for which they need appropriate support. These outcomes were not included in our risk calculator, though they have been the focus of similar work by our colleagues (Meehan et al., 2020).
The valuable insights provided by health and social care professionals and young people will help inform future work related to implementing the risk calculator into practice and, it is hoped, will ultimately improve the outcomes of children exposed to abuse, neglect, domestic violence, and bullying.
These are just the first steps towards that goal, but they are promising ones!
Funding
This research was funded by a research grant from the National Society for Prevention of Cruelty to Children (NSPCC) and the Economic and Social Research Council (ESRC).
Conflicts of interest
The author is an employee of King’s College London and was the lead author of the two research studies presented in this article.
Primary papers
Latham, R.M., Meehan, A.J., Arseneault, L., Stahl, D., Danese, A., and Fisher H.L. (2019), Development of an individualised risk calculator for poor functioning in young people victimised during childhood: A longitudinal cohort study. Child Abuse Negl, 98. doi:10.1016/j.chiabu.2019.104188 [full-text link]
Latham, R.M., Temple, R., Romeo, C., Danese, A., and Fisher, H.L. (2020), Understanding practitioners’ and young people’s views of a risk calculator for future psychopathology and poor functioning in young people victimised during childhood. J Ment Health. doi: 10.1080/09638237.2020.1844869 [Full-text link]
Other references
Cicchetti, D. (2013), Annual research review: Resilient functioning in maltreated children—past, present, and future perspectives. J Child Psychol Psychiatry, 54: 402–422. doi:10.1111/j.1469-7610.2012.02608.x [Full-text link]
Currie, J., and Widom, C. S. (2010), Long-term consequences of child abuse and neglect on adult economic well-being. Child Maltreat, 15: 111–120. doi:10.1177/1077559509355316 [PDF text link]
Herrenkohl, E. C., Herrenkohl, R. C., Egolf, B. P., and Russo, M. J. (1998), The relationship between early maltreatment and teenage parenthood. J Adolesc, 21: 291–303. doi:10.1006/jado.1998.0154 [PDF text link]
Jaffee, S. R., Ambler, A., Merrick, M., Goldman-Mellor, S., Odgers, C. L., Fisher, H. L., Danese, A. and Arseneault, L. (2018), Childhood maltreatment predicts poor economic and educational outcomes in the transition to adulthood. Am J Public Health, 108: 1142–1147. doi:10.2105/AJPH.2018.304587 [Full-text link]
Malvaso, C. G., Delfabbro, P., and Day, A. (2018), The maltreatment-offending association: A systematic review of the methodological features of prospective and
longitudinal studies. Trauma Violence Abuse, 19: 20–34. doi:10.1177/1524838015620820 [Full-text link]
Matthews, T., Danese, A., Gregory, A. M., Caspi, A., Moffitt, T. E., and Arseneault, L. (2017), Sleeping with one eye open: loneliness and sleep quality in young adults. Psychol Med, 47: 2177-2186. doi:10.1017/S0033291717000629 [Full-text link]
Meehan, A. J., Latham, R. M., Arseneault, L., Stahl, D., Fisher, H. L., & Danese, A. (2020). Developing an individualized risk calculator for psychopathology among young people victimized during childhood: A population-representative cohort study. J. Affect. Disord, 262: 90-98. doi: 10.1016/j.jad.2019.10.034 [Full-text link]
Radford, L., Corral, S., Bradley, C., and Fisher, H. L. (2013). The prevalence and impact of child maltreatment and other types of victimization in the UK: Findings from a
population survey of caregivers, children and young people and young adults. Child Abuse Negl, 37: 801–813. doi:10.1016/j.chiabu.2013.02.004 [Full-text link]
