Linden, W. & LeMoult, J. (accepted). Adverse childhood events causally contribute to mental illness: We must act now and intervene early. Journal of Child Psychology and Psychiatry.
Supplement to:
Adverse childhood events causally contribute to mental illness: We must act now and intervene early
By Wolfgang Linden & Joelle LeMoult, Journal of Child Psychology and Psychiatry, 2021
Preamble: The above article was intentionally created as a brief Editorial Perspective. This supplement describes the foundational review that justified the synthesis underlying our viewpoints. The supplement is intended for the reader who wants to ‘dig deeper’ and see more of the evidence.
Review of the criterion-based approach to evidence
Hill’s approach is probabilistic, cumulative and emphasizes plausibility; no absolutes are claimed. While subsequent writers reasserted the usefulness of most of Hill’s criteria, there is fair criticism of the concept of dose-response and specificity (Fedak, Bernal, Capshaw & Gross, 2015; Glass, Goodman, Hern & Samet, 2013). The problem with dose-response is its narrow focus on linear models whereas threshold models or multidimensional models may be equally or even more useful. The specificity criterion is still useful for understanding infectious disease (like CoVid 19) but cannot accommodate the complex, multidimensional mechanisms relevant for ACE sequelae. Therefore, we modified our approach and de-emphasized these two criteria.
This article critically reviews findings about each of Hill’s (Hill, 1965) criteria, modifies and/or reasserts their usefulness as appropriate, and provides a summary of the evidence. We used established systematic reviews where available in conjunction with high-quality empirical work to evaluate causal links and offer policy and intervention recommendations. At the end of this review, a summary table of conclusions is provided as a foundation for the discussion of said recommendations. Note also that Hill’s criteria/viewpoints are not independent of each other. Yet they will be addressed in the traditional order.
Strength
To quantify this criterion, odds or risk ratios are computed and we provide evidence from two pivotal reviews that computed odds/risk ratios by comparing participants with scores of 4 or more ACEs to those with 0-3 ACEs (Hughes, Bellis, Hardcastle, Sethi, Butchart et al., 2017; Felitti, Anda, Nordenberg, GlassWilliamson, Spitz et al., 1998) (see table 1). These two reviews used an overlapping reporting format that allowed direct comparison and they effectively bracket the epidemiological work on ACEs. In 1998, Felitti and colleagues offered data from a very large sample (n=9508) and their study became a major catalyst for two more decades of prospective studies on ACE sequelae. Hughes and colleagues on the other hand systematically aggregated almost two decades’ worth of subsequent prospective work and reported on 37 studies, reflecting 253,719 participants. Felitti and colleagues reported 16 outcome variables that were similarly evaluated by Hughes and colleagues who, however, reported data on 6 additional types of sequelae (see table 1 in the Editorial Perspective).
Hughes and colleagues reported increased risk for all health outcomes for high ACE exposure. Associations were statistically significant but modest for physical inactivity, overweight or obesity, and diabetes (ORs of less than two); fairly large for smoking, heavy alcohol use, poor self-rated health, cancer, heart disease, and respiratory disease (ORs of two to three), and strong for sexual risk taking, mental ill health, and problematic alcohol use (ORs of more than three to six), and strongest for problematic drug use and interpersonal and self-directed violence (ORs of more than seven).
Consistency
Almost every study should support the association between risk factor and outcome for there to be potential causation. The two reviews offering evidence for strength of the relationship between ACE exposure and its sequelae (Table 1) also reveal high consistency (Hughes et al., 2017; Felitti et al., 1998). Note that the ground-breaking Felitti et al. review was published almost 20 years prior but the results are nevertheless strikingly similar and not affected by the passage of time.
Specificity
A risk factor is considered to have specificity when it predicts one particular outcome. When such a ‘pure’ linkage exists, an argument for causality is particularly convincing, but ACEs are by their very nature not narrow and specific nor is there a belief or expectation that ACEs causally contribute to only a singular negative health outcome. Also of note is that ACEs are often comorbid with one another, making it difficult to isolate the specific effect of any one ACE (Glass, Goodman, Hern’an, & Samet, 2013). In fact, as the risk ratios in Table 1 showed, ACE is positively associated with a swath of problematic physical health outcomes, mental illness and high-risk behaviors. As such, one can argue that a risk factor that causally contributes to multiple poor outcomes (in fact almost all known causes of death!) is actually more damaging and more in need of prevention, minimization and/ or treatment than a risk factor with a narrow, singular outcome. Hence, for the case of ACEs, a narrow search for specificity may be contraindicated and underestimate the known, vast and problematic linkages (see Strength and Consistency Criteria above).
Temporality
Putative cause must precede effect and temporality is considered an absolute criterion even if a putative cause has not actually been observed. Showing temporality, namely that ACE preceded ill health is not a problem because all research on ACE is designed to assess ACE before problematic outcomes. A potential problem is that ACE measurement is often retroactive and based on self-report which psychometrics experts are leery of due to recall biases and/or social desirably responding (Baldwin, Reuben, Newbury, & Danese, 2019). On the other hand, typical ACE items are ‘parent in jail’, or ‘family violence’ and are so salient that recall is unusually trustworthy and has led to reliable findings. Furthermore, surveys are typically executed in an anonymous format such that social desirability and impression management are less of a threat to validity (Linden, Paulhus & Dobson et al., 1986). Close attention, however, needs to be given to the length of follow-up when it comes to showing temporal relationships. A 10-year follow-up may suffice to show high delinquency in teenagers with a high ACE background but sequelae like cancer or heart disease will take >50 years to develop and accuracy of recall may have weakened by then.
Biological Gradient
According to the Biological Gradient (or dose-response) criterion, little exposure should result in little effect, and vice versa for extensive exposure. This criterion requires quantifiability of the putative cause. Showing that more exposure to ACEs is associated with progressively worse outcomes is partly already demonstrated in the section on strength and consistency of linkages. Nonetheless, this criterion is difficult to apply because ACEs are multidimensional as well as different in gradient and ACEs are, of course, not connected to just one problematic outcome. In principle, one could compute gradients for each type of ACE with each type of known deleterious outcome but this translates into a herculean task of enormous complexity. In principle, models that use a threshold logic or are multidimensional should also be considered but are rarely tested (Fedak et al., 2015).
Plausibility and Coherence
To be plausible, an outcome should not violate well known laws of the universe; Hill (Hill, 1965) did offer that what is plausible depends upon the biological knowledge of the day. In a related fashion, the criterion of Coherence requires that a cause-and-effect interpretation of data does not seriously conflict with the generally known facts of the biology of a disease. After a struggle to consider these two criteria separately, the authors considered it more parsimonious and logical to integrate discussion of the evidence for these two criteria in a single ‘block.’
Central to determining Plausibility and Coherence is assessing the biological, behavioral and cognitive mechanisms that connect ACEs with mental illness. Reviews of anatomical and physiological changes in various brain areas tend to focus on (a) differential impacts of ACEs on brain development as a function of developmental stage, (b) different biological mechanisms for the impact of threat relative to deprivation, and (c) deleterious outcomes due to toxic stress via changes in biological processes, cognitive function and emotion regulation. The following section provides a very broad description of promising studies on mechanisms.
ACEs may be present for all or most of the first two decades of life and can have deleterious impact when they occur at any age (Lupien, McEwen, Gunnar & Heim, 2009); intervention or prevention may therefore seem equally indicated at any age. However, this article will make a case for the extraordinary importance of a healthy development during pregnancy and the first two years of life (Holtman & Svoboda, 2009) and argues that very early ACE exposure is particularly pernicious.
Some of the most critical stages of brain development begin in the second month of pregnancy and continue until 18 months of life. In the years following, the critical process of pruning excess connections occurs to remove unnecessary neuronal structures from the brain (Ackerman, 1992). Understanding these developmental stages of the nervous system is absolutely critical for formulating effective prevention policies as will be shown later in this article.
In addition to differential synaptic pruning during sensitive periods, ACEs alter neural development through multiple mechanisms (Mclaughlin, Weissman, & Bitran, 2019; Hensch, 2005). Reductions in cortical thickness and volume as well as changes in neural function within these circuits may ultimately contribute to differences in a variety of complex social and cognitive abilities, including language and executive functions (Pollack, Nelson, Schlaak, Roeber, Wewerka, et al., 2010). In addition to these behavioral differences, the dimensional model argues that altered structure and function in the neural networks that support these complex cognitive abilities will also be observed, particularly in the frontoparietal network that supports executive functions (McLaughlin et al., 2019; Sheridan et al. 2014). In a thorough review of brain changes, McLaughlin (McLaughlin, 2016) conclude that some neural abnormalities are associated most consistently with exposure to threat when compared to exposure to deprivation (or mixed threat-deprivation). For example, exposure to threat is associated with reduced amygdala volume, heightened amygdala reactivity to threat cues, reduced volume and thickness of the medial prefrontal cortex, and reduced hippocampal volume.
Comparison of two current models of brain changes as a function of ACE exposure
Table 1 in the printed text displayed risk ratios for ACE outcomes and this type of display represents a cumulative risk model (CR) which sees every adversity as contributing equally to every outcome and as aggregating into a cluster of long-term problems. More recent advances in research on mechanisms do not contradict the usefulness of the CR model but offer a more fine-grained perspective. A proposed dimensional model sees deprivation and threat exposures as differential clusters with certain cluster-specific specific outcomes. Using data from two independent samples, McLaughlin and colleagues (McLaughlin et al., 2019) show a degree of specificity in that this dimensional network theory revealed clusters of adversities and outcomes where threat was associated with emotion reactivity and automatic regulation but deprivation clustered with cognitive outcomes.
A second objective of this section was to place this section on ACE mechanisms within the larger context of research on pathways for stress – disease linkages. We posit that ACE research can be seen as a logical and conceptual extension of almost a century of research on stress and health (Selye, 1936; McEwan & Stellar, 1993) and is consistent with earlier stress research in that both anatomical changes and predisposition for hyper-reactivity of multiple physiological systems are documented in both literatures. High levels of ACE are associated with differential levels of immune markers (Baumeister, Akhtar, Ciufolini, Pariante, & Mondelli, 2016), telomere length (a marker of biological aging)(Chen, Epel, Melon, Lin, Reus, et al., 2014), and are reflected at a cellular level via mitochondrial DNA (Ridout, Khan, & Ridout, 2018). A review of this literature (Gonzalez, Catherine, Boyle, Jack, Atkinson et al., 2018) justifies to call ACEs fitting exemplars of the concept of high allostatic load (McEwan & Stellar, 1993).
These biological changes are mediated by well-documented emotional and cognitive mechanisms which are known to underlie anxiety, depression and personality disorders (McLaughlin et al., 2019; Zeynel & Uzer, 2020; Livesley, 2008; Afsjar-Jalili & Khamseh, 2019). ACEs representing threat predispose the individual to be overly sensitive to fear stimuli and to struggle with emotion regulation (Livesley, 2008) whereas exposure to neglect appears to lead to the development of cognitive schemas characteristic of depression (Afshar-Jalili & Khmaseh, 2019).
Experiments
Evidence is greatly strengthened if a putative relationship is tested under tightly controlled conditions that eliminate, or at least minimize, competing explanations. The randomized controlled trial is the gold standard approach. In the context of ACEs, ethics pose many limits on this otherwise scientifically valuable approach. However, ACE research has been fueled by animal studies that show systematically induced early-life adversity leads to hyper-reactivity of the sympathetic nervous system (SNS) and hypothalamic–pituitary–adrenal (HPA) axis in rodents (Lupien et al., 2009). Although there is limited experimentation in humans, two pivotal studies deserve description (McLaughlin et al., 2016; Kitzman, Olds, Cole, Hnaks, Anson, et al., 2010). McLaughlin and colleagues (2016) had examined whether randomized placement of orphans in Romania into a family caregiving environment alters development when compared to care as usual and to a sample of typically developing Romanian children. Those remaining in institutional care exhibited significantly blunted SNS and HPA axis responses to psychosocial stress compared with children randomized to foster care, whose stress responses approximated those of typically developing children. Importantly, intervention effects were evident for SNS and HPA reactivity only among children placed in foster care before age 24 and 18 months. These findings represent experimental evidence of a sensitive period in humans during which the environment is particularly likely to alter stress response system development.
In the second experimental study, economically disadvantaged children were randomized to a regular public nurse visitation program or a low-intensity active control group that received free transportation to prenatal care, developmental screening and referrals for the child at 6 months, 12 months and 2 years (Kitzman et al., 2010). At the 12-year follow-up children in the home visit group showed less evidence of internalizing disorders, and higher math and reading skills. On the other hand, there was no significant benefit with respect to externalizing disorders or other behavior problems.
Analogy
Using analogies to support an argument for causality cannot be based on quantified data but becomes a question of persuasiveness of the comparison and a broadly-defined fit with known approaches to the science of health. One such analogy comes to mind: tobacco. Tobacco use has a pervasive impact on health outcomes and premature death and plays a demonstrated causative role throughout the full length of life. Tobacco use, especially cigarette smoking can be influenced via public policy, access control, pricing, and population based education. Even smoking differs in many respects from ACE (which does not involve personal choice), the consideration of the struggles, successes and failures to control smoking behavior can teach researchers and policymakers interested in reducing ACE and minimizing its sequelae.
ACE exposure varies in its nature and quantity and so does tobacco use. Type of use can be via cigarettes, vaping, pipe, cigars, or tobacco chewing and length of exposure is linked with ill health outcomes in a gradient fashion. Similar to ACE, some outcomes are relatively specific and associated with a very high risk ratio reaching as high as 50:1 for likelihood of developing lung cancer in heavy long-term smokers (Freedman, Leitsmann, Hollenbeck, Schatzkin & Abnet, 2008). Still relevant outcomes, even though at a much lower level of risk, are cardiovascular and pulmonary diseases.
Tobacco consumption is likely unique because its use can be completely eliminated without any ill consequences. The encouraging news is that despite the challenge of the addictive property of nicotine tobacco use and its sequelae have been drastically cut via government policies, taxation, pricing, access control and public health messaging.
Conclusion from the detailed review of evidence of causality using Hill’s criteria
Unsurprisingly, our review of the literature repeats the finding that ACE exposure strongly correlates with a wide swath of deleterious long-term outcomes. Notably, ACEs have more impact on mental illness and problematic behavior outcomes than on physical illnesses (which, however, are still far from negligible and may take much longer to become evident disease). ACEs alone cannot explain all mental illness but deserve attention because they offer the potential for prevention, modification, or at least minimization.
References for the Supplemental Text
Ackerman S. (1992). Discovering the Brain. Washington (DC): National Academies Press (US)
Afshar-Jalili Y., Khamseh A. (2019). Why childhood toxic experiences and early maladaptive schemas affect negatively one’s psychological capital. Journal of Psychology and Psychotherapy, 2019; 9: 365. DOI 10.35248/2161-487.19.9.365
Baumeister D., Akhtar S., Ciufolini S., Pariante C.M. & Mondelli V. (2016). Childhood trauma and adulthood inflammation: a meta-analysis of peripheral C-reactive protein, interleukin-6 and tumour necrosis factor-α. Molecular Psychiatry, 21: 642–649
Bethell C.D., Carle A., Hudziak J., Gombojav N., Powers K., Wade R., Braveman P. (2017). Methods to Assess Adverse Childhood Experiences of Children and Families: Toward Approaches to Promote Child Well-being in Policy and Practice. Academic Pediatrics, 17: S51–S69
Chen S.H., Epel E.S., Mellon S.H., Lin J., Reus V.I., Rosser R., Kupferman E., Burke H., Mahan L., Blackburn L.H., Wolkowitz O.M. (2014) Adverse childhood experiences and leukocyte telomere maintenance in depressed and healthy adults. Journal of Affective Disorders, 169: 86–90.
Fedak K.M., Bernal A., Capshaw Z.A., Gross S.(2015). Applying the Bradford Hill criteria in the 21st century: how data integration has changed causal inference in molecular epidemiology. Emerging Themes in Epidemiology, 12:14. DOI 10.1186/s12982-015-0037-4
Felitti V.J., Anda R.F., Nordenberg D., Williamson D.F., Spitz A.M., Edwards V., Marks J.S. (1998). Relationship of childhood abuse and household dysfunction to many of the leading causes of death in adults: The Adverse Childhood Experience (ACE) Study. American Journal of Preventive Medicine, 14: 245-258.
Freedman N.D., Leitzmann M.F., Hollenbeck A.R., Schatzkin A., Abnet C.C. (2008) Cigarette smoking and the subsequent risk of lung carcinoma in the men and women of a large prospective cohort study. Lancet Oncology, 9: 649–656. doi: 10.1016/S1470-2045(08)70154-2
Glass T.A., Goodman S.N., Hern´an M.A., Samet J.M. (2013). Causal Inference in Public Health. Annual Review Public Health, 34:61–75
Gonzalez A., Catherine N., Boyle M., Jack S.M., Atkinson L., Kobor M., Sheehan D., Tonmir L., Waddell C., Macmillan H.L., on behalf of the Healthy Foundations Study Team. Healthy Foundations Study (2018). A randomised controlled trial to evaluate biological embedding of early-life experiences. BMJ Open, 8:e018915. doi: 10.1136/bmjopen-2017-018915
Hensch T. (2005). Critical period plasticity in local cortical circuits. Nature Review Neurosciences, 6:877–888
Hill A.B. (1965). The Environment and Disease: Association or Causation? Proceedings of the Royal Society of Medicine, 58: 295–300. doi:10.1177/003591576505800503
Holtmaat A., Svoboda K. (2009). Experience-dependent structural synaptic plasticity in the mammalian brain. Nature Review Neurosciences, 10: 647–658
Hughes K., Bellis M.A., Hardcastle K.A., Sethi D., Butchart A., Mikton C., Jones L. , Dunne M.P. (2017) The effect of multiple adverse childhood experiences on health: a systematic review and meta-analysis. Lancet Public Health, 2: e356–e366
Kitzman H.J., Olds D.L., Cole R.E., Hanks C.A., Anson E.A., Arcoleo K.J., Luckey D.W., Knudtson M.D., Henderson C.R., Holmberg J.R. (2010). Enduring Effects of Prenatal and Infancy Home Visiting by Nurses on Children: Follow-up of a Randomized Trial Among Children at Age 12 Years. Archives Pediatric and Adolescent Medicine, 164: 412-418. doi:10.1001/archpediatrics.2010.76
Linden W., Paulhus D.L., & Dobson K.S. (1986). Effects of response styles on the report of psychological and somatic symptoms. Journal of Consulting and Clinical Psychology, 54: 309-313
Livesley J. (2008). Toward a genetically-informed model of borderline personality disorder. Journal of Personality Disorders, 22: 42-71
Lupien S.J., McEwen B.S., Gunnar M.R., Heim C. (2009). Effects of stress throughout the lifespan on the brain, behaviour and cognition. Nature Reviews, 10: 434-445
McEwen B.S., Stellar E. (1993). Stress and the individual: Mechanisms leading to disease. Archives of Internal Medicine, 153: 2093-2101.
McLaughlin K.A., Sheridan M.A., Tibu F., Fox N.A., Zeanah C.H., Nelson C.A. (2016). Causal effects of the early caregiving environment on stress response system development in children. Proceedings of the National Academy of Science, 112: 5637-5642.
McLaughlin K.A., Sheridan M.A.(2016). Beyond cumulative risk: A dimensional approach to childhood adversity. Current Directions in Psychological Science, 25: 239-245.
McLaughlin K.A., Weissman D., Bitrán D. (2019). Childhood Adversity and Neural Development: A Systematic Review. Annual Review of Developmental Psychology, 1: 277–312.
McLaughlin K.A. (2016). Future directions in childhood adversity and youth psychopathology. Journal of Clinical Child and Adolescent Psychology, 45: 361-382.
Pollak S.D., Nelson C.A., Schlaak M.F., Roeber B.J., Wewerka S.S., Wiik K.L., Frenn K.A., Loman M.M., Gunnar M.R . (2010) Neurodevelopmental effects of early deprivation in post-institutionalized children. Child Development, 81: 224–236
Ridout K.R., Khan M., and Ridout S.J. (2018) . Adverse Childhood Experiences Run Deep: Toxic Early Life Stress, Telomeres, and Mitochondrial DNA Copy Number, the Biological Markers of Cumulative Stress. BioEssays, 40: 1800077
Selye H. (1936). A syndrome produced by diverse nocuous agents. Nature, 138.
Zeynel Z., Uzer T. (2020). Adverse childhood experiences led to trans-generational transmission of early maladaptive schemas. Child Abuse and Neglect, 99: 104235.