Dr Umar Toseeb Hello, welcome to the Podcast Series for the Association for Child and Adolescent Mental Health, or ACAMH for short. I’m Umar Toseeb, Professor of Psychology. My research focuses on special educational needs and mental health in childhood and adolescence. All listeners to this, and indeed any of ACAMH’s podcasts, are eligible for a free CPD certificate. Do please visit acamhlearn.org for details of this, together with information on how you can access hundreds of hours of free talks, lectures, interviews, all of which you can also get free CPD certificates for. The web address is acamhlearn.org, that’s a-c-a-m-h-l-e-a-r-n.org. Today, I’ll be speaking to Athena Chow, Lead Author of the paper, “Do Meaningful Dimensions of Childhood Adversity Exist? Data-driven Evidence from Two Prospective Cohort Studies,” published in the JCPP, the paper will be the focus of today’s podcast. Athena, thank you so much for joining me. Athena Chow Thank you so much for having me. Dr Umar Toseeb Can we start with some introductions? Can you tell us about what you do and what your research interests are? Athena Chow Hi, Athena Chow, a final year PhD candidate in experimental psychology at the University of Oxford. My PhD research focuses on investigating the relationship between adverse childhood experiences and adolescent psychopathology, and my current work forms part of the Biopsychosocial Doctoral Programme funded by the Leverhulme Trust. I’m particularly interested in using data-driven methods and large longitudinal dataset to study how childhood adversity increases the risk for mental health problems over time. Dr Umar Toseeb Thank you, and who are some of the Co-Authors on the paper? Athena Chow My Co-Authors are Professor Lucy Bowes and Dr Jessie Baldwin, and they are my PhD Supervisors, whom I’m very lucky to work with. So, Lucy is a Professor of Developmental Psychopathology at Oxford, who leads the oRANGE Lab, where we integrate methods from epidemiology, experimental studies and evidence-based interventions, to understand the trajectories of risk and resilience. And Jessie is a Senior Research Fellow at UCL, who has mentored me since I was an undergrad, and she specialises in using causal inference methods to investigate the extent to which childhood adversity causes mental health problems. Dr Umar Toseeb Thank you, and before we go onto talk about the paper itself, I think it might be helpful for us to go through some definitions. So, the paper is about “adverse childhood experiences,” or “ACEs,” can you just give us some background of the origins of that term, and what it means? Athena Chow ACEs are stressful and traumatic experiences that occur during childhood and can have a lasting impact into adolescence, adulthood, and even across the lifespan. And perhaps the well-known study is the CDC-Kaiser ACE Study which classified ACEs as “exposures to multiple types of abuse, neglect and domestic violence, and other types of household dysfunction, like substance abuse and caregiver mental illnesses.” So, these ACEs were then summed as a risk score, to show a dose response relationship. Basically, the more ACEs a person experienced, the higher their risk was for negative health outcomes later on in life. And the ACE Study showed that the more ACEs you experienced, the higher risk you were at for mental health problems, like depression and anxiety, and chronic diseases, such as cancer and heart disease, and even alcoholism and drug abuse. Dr Umar Toseeb I think many people would have heard of ACEs, and then the ten original ACEs that you’ve just talked about, but then I think in your introduction, you talk – also talk about the ‘extended ACEs’, and the more diverse range of adverse childhood experiences. Do you just want to expand on that for us? Athena Chow As the field has moved forward, research has expanded the original ten ACEs to include more diverse experiences, like peer victimisation, community violence and low socioeconomic status, which have all been shown to significantly improve the prediction of mental and physical health problems. And the evidence has shown that broadening the definition of ‘adversity’ beyond the ten original ACEs could help us to advance our understanding of the relationship between childhood adversity and psychopathology. Dr Umar Toseeb And then in your discussion of the ACE model, you talk about the “cumulative approach to ACEs,” and you’ve just mentioned it a few minutes ago. What is the cumulative approach to adverse childhood experiences, and why is it limited? Athena Chow The cumulative risk approach is an approach that sums the number of ACEs experienced as a total ACE score. So, each adverse experience is assigned a point and the total score represents the cumulative risk. This approach focuses on the number of adversities, rather than the type of adversity or the severity of the adversity. to put it simply, it focuses on quantity over quality, and the main limitation of this approach is that it treats all ACEs as equally harmful, and it assumes that each additional adverse experience contributes equally to the overall risk of negative outcomes. But we know that this is not true, because two children with the same ACE score could have experienced very different types of adversities with very different consequences. For example, research has shown that sexual abuse and parental divorce impact physical and mental health in distinct ways, but the cumulative risk approach would sum these very different experiences together and treat them equivalently. Dr Umar Toseeb Yeah, so I think in that example, so parental separation and sexual abuse, so I think what you’re saying is both of those children, if one’s experienced one and the other one’s experienced the other, they would both get a score of one, but the relationship with subsequent psychopathology would be different, and the cumulative approach doesn’t take that into account. Is that what you’re saying? Athena Chow Yes, exactly, yeah. Dr Umar Toseeb And then, so what is the alternative to that? So, you know, I think the advantage of having an adverse childhood experiences cumulative approach is that it’s very simple to administer and calculate, once you’re doing the research, and it’s also quite easy to understand when you are trying to communicate it to other Researchers, but other stakeholders who aren’t necessarily Researchers. So, what is an alternative to that? Athena Chow Yeah, so there are definitely advantages of the cumulative risk approach and as you said, it provides a really intuitive tool that has been very useful in research and practice. But to overcome the limitations of the cumulative risk model, recent research has started to use dimensional models. And the Dimensional Model of Adversity in Psychopathology, also known as DMAP, this model proposes that ACEs can be conceptualised along two core dimensions, threat and deprivation, and each dimension impacts mental health through distinct mechanistic pathways. So, DMAP defines “threat” as “experiences of physical or emotional harm, such as physical, sexual and emotional abuse.” And “deprivation” is defined as “the absence of expected inputs from the environment, such as cognitive stimulation and emotional stimulation.” So, the experience of physical or emotional neglect would be classified as “deprivation.” Dr Umar Toseeb So, is it still, like, a sum type approach, where you’re adding up the number of experience – adverse experiences, and a difference is that you’re not clumping them all – all ten of them together, for example. You’re now taking a dimensional approach and working on the assumption that the experiences that fall within the threat dimension contribute approximately equal to psychopathology, and then the ones that are in the deprivation dimension contribute approximately equal to psychopathology? Athena Chow Yeah, so that’s a great question. I don’t think that the dimensional approach assumes that experiences within each dimension contribute equally to the development of psychopathology. But rather I think conceptualising them as dimensions, it – by doing that, it assumes that each dimension will work through the similar mechanism and a similar pathway. And this might be a neurobiological or cognitive mechanism, like emotional regulation or executive function, and then these distinct pathways, per dimension, can then lead to different developmental consequences. Dr Umar Toseeb Very helpful, and as with the cumulative approach to ACEs, I imagine there are some drawbacks of a dimensional model. What are those drawbacks? Athena Chow Yeah, so the drawbacks, I would say they’re not – the issue isn’t to do with the two dimensions per se, but rather the methods used to derive the two dimensions. So, subjective practices in defining ACE measures can lead to inconsistent dimensions of adversity across studies, because each study might choose different measures for the same construct. In our paper, we discuss a case where two independent studies derived a different number of dimensions, even though they both used data from the same longitudinal cohort study, and they had both also used DMAP to categorise their dimensions. So, although DMAP has provided a really useful framework to conceptualise ACEs as dimensions, most studies have used confirmatory approaches, like confirmatory factor analysis. But the issue with confirmatory factor analysis is that it relies on predefined categorisations of ACEs. Dr Umar Toseeb There was a paper, a research review, in the JCPP, towards the end of last year, where I think your Supervisor, Jessie Baldwin, was a Co-Author on that, where they talked about why maltreatment measures that are prospective versus retrospective differ… Athena Chow Yeah. Dr Umar Toseeb …and they touch on some of the issues that you’ve just talked about. So, I think it’s definitely an issue that is out there and that is getting some traction, which is fantastic. So, based on what we’ve talked about so far, so we’ve talked about this “cumulative approach to ACEs,” and then we’ve talked about a “dimensional approach to ACEs,” what is the gap in the literature that your work is trying to address? Athena Chow I think the gap that our study is trying to address is the approach to deriving the dimensions. So, because previous research has largely relied on confirmatory approaches, they can be biased by predefined categories of ACEs, which then results in inconsistent dimensions derived across studies. So, to address this gap, we instead, decided to use a data-driven exploratory approach, which means that with the exploratory factor analysis we applied, we allowed each ACE to load onto different dimensions, in different numbers, in different combinations, without specifying any predefined structure. And then we chose the best fitting model according to the best fitting statistics. So, essentially, we let the data speak for itself. And to our knowledge, our study is the largest pre-registered data-driven analysis of ACE dimensions, using two contemporary longitudinal cohorts from the UK and the US. Dr Umar Toseeb And what were those two datasets? Athena Chow We used the UK Millenium Cohort Study, which follows over 18,000 British children born between the year 2000 and 2002, which I know that you have also lots of experience using this dataset. And the other dataset we used was the American Adolescent Brain Cognitive Development, ABCD Study, and this cohort follows over 11,000 American children born between the year 2006 and 2008, across the United States. Dr Umar Toseeb And I know that in each of the datasets, and this is just the case when you use lots of different datasets, the measures would be slightly different. So, do you just want to tell us what your measures of adverse childhood experiences were in the two datasets, and then, also, what your measures of psychopathology were in the two datasets? Athena Chow We identified a range of ACEs in both the MCS and the ABCD, and these ACEs included the original ten ACEs, such as physical, sexual and emotional abuse, emotional neglect, domestic violence and parental mental health problems. We also included more diverse ACEs, like peer victimisation, community violence, neighbourhood deprivation and bereavement. And of course, the measures used from each dataset would be slightly different, but wherever it was possible, we derived equivalent ACEs across the two datasets, and this resulted in 18 composite measures of ACEs for each dataset. In terms of the outcomes, we chose to use internalising and externalising symptoms of psychopathology in the MCS and the ABCD. So, in the MCS, we measured this using child self-reports from the Strengths and Difficulties Questionnaire, and in the ABCD, we used parent reports from the Child Behavior Checklist. Dr Umar Toseeb And what age were these measures taken? So, I suppose there was a range, I imagine… Athena Chow Hmmm. Dr Umar Toseeb …with the adverse childhood experiences, what was that age range? And then what was the age or age range for the psychopathology measures? Athena Chow Yeah, so the age range for the ACE measures in the MCS, we derived them across nine months to the age of 14, and then with the ABCD, we derived them across birth to age 11 to 12 years. So, with the ABCD, that was actually the most recent sweep of data that was available at the time when we accessed the data. In terms of psychopathology, in the MCS, we measured the symptoms at age 17, and with the ABCD, we measured it at ages 12 to 13. Dr Umar Toseeb Thank you. So, you’ve got the adverse childhood experiences measures in early life to early adolescence-ish, and then you’ve got your psychopathology measures that come afterwards. What did you find? Athena Chow We found that in the MCS, ACEs loaded onto four dimensions of parental threat, deprivation, victimisation and parental discipline. So, the parental threat dimension included ACEs like parental drug and alcohol use and domestic violence. In the deprivation dimension, this included ACEs like low household income and neighbourhood deprivation. Then the victimisation dimension consisted of physical, verbal, sexual and peer victimisation, and finally, for the parental discipline dimension, this included parental smacking and harsh parental discipline. In the ABCD we found that ACEs loaded onto four dimensions of parental threat, deprivation, victimisation and traumatic events. So, the first three dimensions were largely consistent with what we found in the MCS, and then the fourth dimension, the traumatic events dimension, this included experiences like domestic violence, physical, emotional and sexual abuse, as well as, community violence and bereavement. Dr Umar Toseeb And just remind us, so you said, ‘the first three’, which three dimensions were consistent across the datasets? Just so we can get our head round this. Athena Chow So, we found that the first three dimensions of parental threat, deprivation and victimisation, both emerged consistently in both the MCS and ABCD cohorts. So, the consistency of these three dimensions suggests that they were meaningful and not sample specific. Because they replicated even though there were sociodemographic differences between the UK and the US populations, and, also, despite the fact that ACEs in each cohort were measured using different instruments and different informants and at slightly different ages. Dr Umar Toseeb And what are some of the example indicators of each of the dimensions? So, like, within victimisation, for example, what kinds of things are being measured? Athena Chow So, for example, in the ABCD, the victimisation dimension included peer and cyber victimisation. In the MCS, it was different forms of victimisation, like physical, verbal and sexual victimisation. And then in terms of the parental threat dimension, both cohorts included ACEs, like parental substance abuse, and then for the deprivation dimension, both cohorts included ACEs, like neighbourhood deprivation, lack of neighbourhood safety and low household income. Dr Umar Toseeb Excellent. So, you’ve taken these adverse childhood experiences and you’ve taken a data-driven approach to see whether there are dimensions, and your rationale for doing that was that potentially, the different dimensions, if they exist, might have different mechanisms or pathways to psychopathology. So, did you find that? Athena Chow We found that for both populations, dimensions had differential effects on psychopathology. So, victimisation displayed the strongest associations with psychopathology, over and above the other dimensions. And then the dimension of parental threat was consistently associated with increased risk for internalising and externalising symptoms, across both cohorts. And the dimensions of parental discipline and traumatic events also demonstrated smaller but consistent associations with psychopathology. However, although deprivation did emerge as a meaningful dimension, there was no association between deprivation and psychopathology in the MCS, and very small associations between deprivation and psychopathology in the ABCD. These findings support our hypothesis that threat would be more strongly associated with psychopathology than deprivation, and they actually do align with recent ABCD studies, which found that the dimension of deprivation displayed no or minimal associations with psychopathology. So, this is not to say that deprivation has no association with psychopathology. Instead, we think that the impact of deprivation on psychopathology appears to be inconsistent in the samples, because deprivation is a complex construct that is multi-dimensional in itself. Dr Umar Toseeb I think what you’re saying is it’s not that deprivation isn’t associated with psychopathology, it’s that… Athena Chow Yeah. Dr Umar Toseeb …measurement of deprivation, or how it’s conceptualised in different studies, is different between studies, and that might affect its relationship with psychopathology, or the observed relationship with psychopathology? Athena Chow Yeah, definitely. So, previous research has shown that the strength and the consistency of the association between deprivation and psychopathology, it can vary according to the dimension that is used. So, for example, individual versus material deprivation, or subjective versus objective indicators of deprivation. And then, again, this brings us back to the paper you mentioned earlier that Jessie had done on objective versus subjective measures of ACEs. So, I think this could be an avenue for future research to investigate which dimensions of deprivation do influence psychopathology. Dr Umar Toseeb And in your analysis, did you do anything on sex differences, and if so, did you find anything? Athena Chow So, we also pre-registered some sensitivity analyses to investigate whether sex interacted with adversity to influence psychopathology, because we were informed by some previous research, which did find some interactions. So, in our study, we did find small, significant interactions. Specifically, in the MCS, we found that girls who experienced victimisation were at slightly higher risk for internalising symptoms than boys. Whereas in the ABCD, boys who experienced parental threat were at slightly higher risk for externalising symptoms than girls. However, these results did not replicate across the cohorts, and we did not find sex by adversity interactions for most of the dimensions and psychopathology outcomes. So, I think this could also be another avenue for future research, for Researchers who want to look at these sex differences in more detail, and we do report the details of our sensitivity analysis in our supplement. Dr Umar Toseeb Why would you expect a difference in the relationship between adverse childhood experiences and psychopathology for boys and then for girls? Like, what is it about their experiences, or about being a boy, or treatment of boys, or treatment of girls, or being a girl, that would affect that relationship? Athena Chow Yeah, that’s a great question. I think from the pattern we observed, where girls who experienced victimisation were at slightly higher risk for internalising symptoms than boys, that’s definitely something that would be really interesting to study in more detail. Because I think girls might be at higher risk for internalising symptoms anyway, compared to boys. Like, there’s definitely the gender gap in depression and anxiety. Like, we know the statistics, girls are just at a significantly higher risk for depression and anxiety, especially at this age. And in the ABCD, we also did observe the link where they were at slightly higher risk for externalising symptoms if they had experienced parental threat. And again, we do know the statistics for boys being at higher risk for externalising disorders, compared to girls. So, I think there’s definitely a lot that could be delved into further for this research area. Dr Umar Toseeb Could it be that the reason why you’ve observed some inconsistent findings between the cohorts is because your measure of psychopathology was self-report in one dataset, and then parent report in the other dataset? And I’m sure you’re aware they’ve seen other studies where people argue that depending on the source of reporting for psychopathology, you’re measuring, potentially, a slightly different thing. Because young people themselves might be very good at reporting internalising symptoms, in terms of what they’re experiencing and how they are feeling, and parents might be quite good at reporting behavioural problems or things that manifest externally. So, could that explain why you found a difference between the datasets? Athena Chow I think so, yeah, that’s definitely a potential explanation for why we had these inconsistencies and differences between the datasets. And I think this, kind of, then brings us to the issue of shared method variance and the different ways that, you know, self-reports and parent reports can affect our analyses. And I’m sure you know this very well, that shared method variance is just a caveat that all Researchers who work with secondary datasets have no choice but to deal with, because when you work with secondary data, you are ultimately constrained by the measuring instruments that the study has used. Dr Umar Toseeb It is really interesting, so I just submitted a paper where it is an issue, and I was, like, torn between whether we should highlight it as a problem in the limitations, just don’t mention it, just ‘cause I was, like, everyone knows this is a problem, and do we need to keep mentioning it? So, I was, like… Athena Chow Yeah. Dr Umar Toseeb …I don’t know whether I should say it or not, because it’s just a reason for someone to reject this paper. I was, like – I’m, like, “Oh, I don’t know,” but anyway… Athena Chow Yeah. Dr Umar Toseeb …fingers crossed, it’s going to be fine. Athena Chow Yes, I’m sure it will be, yeah. Dr Umar Toseeb So, in terms of what these findings mean for other Researchers, what would you – what do you think are the implications of your work for other Researchers, but also for people who might work with children and young people? Athena Chow Our findings provide empirical support for the dimensional approach of conceptualising ACEs. So, this shows the advantages the dimensional approach has over the cumulative risk approach, and I think for Researchers and Practitioners, it shows that different types of ACEs clearly impact adolescent mental health in different ways, and so they should not be lumped together into a cumulative risk score. And on the research side, we’ve demonstrated that applying a data-driven exploratory approach, without predefined categorisations, can successfully capture meaningful dimensions of childhood adversity across two distinct populations. And so, I think this also, kind of, highlights the importance of having a clear pre-registered protocol, that you can then replicate across two different cohorts. And while our dimensions of parental threat and deprivation do provide partial support for DMAP, our findings also suggest that existing frameworks of ACEs should be expanded to include victimisation as a distinct dimension, because we found in our study that victimisation was most strongly associated with psychopathology over and above the other dimensions. So, future research could definitely investigate whether victimisation ACEs impact psychopathology via different mechanisms, and it could also be something to keep in mind for policymaking, in that the interpersonal nature of victimisation, particularly when the perpetrator is a peer, it might harm adolescents’ mental health in a more targeted way, compared to other less relational ACEs. Dr Umar Toseeb Yeah, I can definitely see that being the case, and yeah, I had a master’s student who did his dissertation on “Adverse Childhood Experiences and Their Relationship with Subsequent Psychopathology in Adolescence.” And then I think school connectedness, as in their sense of belonging and feeling like they connect to a school was a – to their school, was, like, a moderate or immediate, I don’t know which one. But I imagine that relationship would differ depending on the type of ACE. As you say, if it’s relational, I would imagine it’d affect a young person’s sense of belonging and connectedness to a school more than it might if it was a different type of ACE. So, I can definitely see that being a fruitful avenue for future research. And you said you were a final PhD year student, so what have you got coming up that we should look out for in this area? Athena Chow Thanks for that question. I’m currently working on some exciting analyses for the final chapter of my PhD thesis, where I’m using causal inference methods, specifically propensity score matching, to estimate the causal effects of ACEs on the risk for depression and anxiety. But I think my main priority right now is just to finish my thesis and survive the PhD. Dr Umar Toseeb Okay, and finally, what’s your take home message for our listeners? Athena Chow So, the first take home message is that meaningful dimensions of ACEs do exist, and they can replicate across populations if data-driven methods and consistent protocols are used. Our paper provides an open science resource of ACE measures, many of which have not yet been derived in the MCS and ABCD Study. So, we hope that our code will be replicated by Researchers in future studies, and that this can contribute to the reproducibility of ACE research in both the MCS and the ABCD. The second take home message is victimisation should be considered an adverse childhood experience. Many recent studies still do not include peer victimisation as an ACE, likely because it was not one of the original ten ACEs from the CDC-Kaiser ACE Study. Yet, we found in our study that victimisation displayed the strongest link to psychopathology symptoms in adolescents. So, we can see that ACEs like peer victimisation might impair adolescent mental health in a really targeted way, and this highlights the need for policymakers to develop more targeted programmes, like anti-bullying initiatives, that can address the most tractable forms of childhood adversity. Dr Umar Toseeb Thank you, Athena, for that, and thank you for taking the time to speak to us. For more details, please visit the ACAMH website, www.acamh.org. ACAMH is spelt A-C-A-M-H. And don’t forget to follow us on your preferred streaming platform, let us know if you enjoy the podcast, with a rating or review, and do share with your friends and colleagues.