Abstract
Childhood abuse, neglect, and loss are common in psychiatric patients, and the relationship between childhood adversity and adult mental illness is well known. However, beyond diagnoses that are specifically trauma-related, such as posttraumatic stress disorder, there has been little research on how childhood adversity contributes to complex presentations that require more intensive treatment. We examined the relationship between childhood adversity and other contributors to clinical complexity in adult outpatients seeking mental health assessment. In a cross-sectional study, patients completed standard measures of psychological distress and functional impairment. Psychiatrists completed an inventory of clinical complexity, which included childhood abuse, neglect, and loss. Of 4,903 patients seen over 15 months, 1,315 (27%) both consented to research and had the measure of complexity completed. Childhood abuse or neglect was identified in 474 (36.0%) and significant childhood loss in 236 (17.9%). Correcting for multiple comparisons and controlling for psychiatric diagnosis, age, and sex, patients with childhood abuse or neglect were significantly more likely to also have 11 of 31 other indices of clinical complexity, with odds ratios ranging from 1.7 to 5.0. Both childhood abuse or neglect and childhood loss were associated with greater overall complexity (i.e., more indices of complexity, χ2 = 136 and 38 respectively, each p < .001). Childhood abuse and neglect (but not childhood loss) were significantly associated with psychological distress (Kessler Psychological Distress Scale [K10] score, F = 6.2, p = .01) and disability (World Health Organization Disability Assessment Scale 2.0 [WHODAS 2.0] score, F = 5.0, p = .03). Childhood abuse and neglect were associated with many characteristics that contribute to clinical complexity, and thus to suboptimal outcomes to standard, guideline-based care. Screening may alert psychiatrists to the need for intensive, patient-centered, and trauma-informed treatments. Identifying childhood adversity as a common antecedent of complexity may facilitate developing transdiagnostic programs that specifically target sources of complexity.
Keywords
Childhood experience with abuse, neglect, or loss increases the risk of many adult mental illnesses (Afifi et al., 2014; Lindert et al., 2014; Springer, Sheridan, Kuo, & Carnes, 2007), well beyond specifically trauma-related diagnoses such as posttraumatic stress disorder (PTSD). As a result, childhood adversity is a common consideration in adult psychiatric care. Although the link between childhood adversity and mental illness is well-established, the relationship between childhood adversity and complex presentations of mental illness has received less study.
Complexity in medical settings is difficult to define and to measure (Katerndahl, Parchman, & Wood, 2009). Although complexity in health care has been defined in various ways (Kannampallil, Schauer, Cohen, & Patel, 2011), one definition that highlights its clinical importance is that it refers to “interference with the achievement of expected or desired health and service use outcomes . . . due to the interaction of biological, psychological, social and health system factors” (Lobo et al., 2015). While biological, psychological, and social factors are important for every illness, this definition identifies complex presentations as those in which the interaction of these factors interferes with outcomes when patients receive standard care. Indeed, complexity interferes with many outcomes, contributing to, for example, higher costs, more complications, more hospitalizations, and decreased quality of life for patients (Fortin et al., 2006; Librero, Peiro, & Ordinana, 1999). The Cynefin framework is useful in this context, because it links complexity to the challenges of applying treatment guidelines, which usually focus on single diagnoses without consideration of biopsychosocial interactions. The Cynefin framework classifies problems as either “obvious” (easily identified and treated), “complicated” (problems whose causes and consequences can be discerned by experts), “complex” (problems with unpredictable and changeable causes and with consequences which can only be known in retrospect), or “chaotic” (problems too urgent and changeable to allow for reflection on causes and consequences) (Van Beurden, Kia, Zask, Dietrich, & Rose, 2013). Within this framework, treatment guidelines are essential for obvious problems and useful for complicated ones, but often fail to apply to problems that are complex or chaotic. Primary care providers for complex patients with multidimensional needs, including socioeconomic, medical, and mental health problems, describe optimal care as emphasizing aspects of care that do not typically emerge from treatment guidelines: coordinating care, preventing hospitalizations, and developing patient trust using a team-based approach (Loeb, Bayliss, Candrian, deGruy, & Binswanger, 2016).
Clinical complexity is important in psychiatry because social, psychological, and biological contributors to complexity are so common in this context. Indeed, the Diagnostic and Statistical Manual of Mental Disorders (5th ed.; DSM-5; American Psychiatric Association, 2013) indicates that “the case formulation for any given patient must involve a . . . concise summary of the social, psychological and biological factors that may have contributed to developing a given mental disorder” (p. 19). Following the definition provided above, we understand that these factors interact in ways that would compromise treatment outcomes if patients received standard care (e.g., care that follows treatment guidelines for a single diagnosis). A data-driven approach to identifying indices of complexity confirmed that patients identified as having greater complexity have poorer outcomes with standard psychological treatment but could do better if assigned to more intensive treatment (Delgadillo, Huey, Bennett, & McMillan, 2017). The World Psychiatric Association has advocated that clinical complexity in psychiatry should be met with a person-centered, integrative approach to diagnosis that is based on a bio-psycho-socio-cultural framework recognizing multiple dimensions of complexity (Mezzich & Salloum, 2008). This is challenging because complexity emerges from factors that interact in many ways.
Among many reasons that contributors to complexity are interrelated, some factors may share a common antecedent. Childhood adversity may be such a “cause of causes.” Types of childhood adversity that influence later health include physical, emotional, and sexual abuse; various forms of neglect; and exposure to household violence and conflict, bullying, poverty, and parents with mental illness or addictions (Felitti et al., 1998; Finkelhor, Shattuck, Turner, & Hamby, 2013). Childhood adversity is related to a number of social determinants of health that are likely to make living with mental illness more challenging and treating it more complex. These include lower education, greater unemployment, and lower income (Currie & Widom, 2010; Freisthler, Merritt, & LaScala, 2006). Childhood adversity is also associated with other characteristics that complicate psychiatric care, such as chronic pain (Davis, Luecken, & Zautra, 2005) and personality disorders (Spataro, Mullen, Burgess, Wells, & Moss, 2004). Thus, experience with childhood adversity both increases the risk of many mental illnesses and increases the likelihood of a range of environmental and personal characteristics that could interact to make the treatment of mental illness more complex.
The evidence for the relationship between early adversity and factors that contribute to complexity, reviewed above, is derived from studies of individual factors rather than studies of multiple interacting factors, or complexity itself. The literature addressing the relationship between childhood adversity and complexity, in this sense, is small but includes a study of psychiatric outpatients for whom multiplicity of symptoms (as measured with the Minnesota Multiphasic Personality Inventory–2 [MMPI-2]) was greater for those with a history of child abuse (Choi, Choi, Gim, Park, & Park, 2014). With respect to complexity in medicine, in a population-based cohort childhood adversity was associated with increased odds of multimorbidity of chronic diseases (Sinnott, Mc Hugh, Fitzgerald, Bradley, & Kearney, 2015).
Thus, existing evidence suggests that childhood adversity is linked to complexity in psychiatry, but there has been little study of the association between childhood adversity and multiple, potentially interacting, contributors to complex presentations. This is important because if childhood adversity is linked to many concurrent contributors to the interactions that create clinical complexity, then screening for childhood adversity can serve to alert psychiatrists to the likelihood that more intensive, patient-centered, and trauma-informed treatments may be necessary. The current study uses a recently developed psychiatric complexity inventory to address this gap. This tool, the C4 Inventory, takes advantage of the fact that multiple contextual factors are often assessed clinically during a psychiatric assessment interview, and uses a checklist-format inventory to systematically record their presence or absence after a clinical assessment (Maunder, Wiesenfeld, Rawkins, & Park, 2016). The purpose of this article is to examine the relationship between a clinically relevant history of childhood adversity as identified in the C4 Inventory and other contributors to clinical complexity in a real-world sample of outpatients seeking mental health assessment.
Method
Procedure and Participants
This study examined relationships among measures gathered at a single point in time. Participants were outpatients referred for psychiatric assessment to psychiatrists in an academic general hospital. At the time of first assessment in this department, standard measures are often completed by patients (psychological distress, functional impairment) and psychiatrists (clinical complexity, overall function) and are entered into a database along with diagnostic and demographic data. Patients may choose to consent to this information being used for research purposes or not. This process was approved by the Mount Sinai Hospital Research Ethics Board.
Of patients seen for outpatient assessment between January 2014 and March 2018 for whom any standard measures were collected (n = 4,903), 2,259 (45.9%) consented to these measures being used for research. Among consenting patients, the C4 Inventory had been completed for 1,315 patients (1,315/2,259 = 58.2%, 1315/4903 = 26.8%), who form the cohort for this analysis. As a clinician-completed instrument, whether or not the C4 Inventory is completed is primarily related to the logistics of a busy outpatient department. Among patients who consented to research, those for whom C4 Inventory data were available were somewhat younger (mean age 42.5 ± standard deviation 14.5 years vs. 46.4 ± 21.3, p < .001), less likely to be female (50% vs. 56%, p < .001), and reported somewhat more psychological distress (Kessler Psychological Distress Scale [K10] mean score, 28.5 ± 10.4 vs. 26.5 ± 11.6, p < .001). They did not differ in functional impairment (World Health Organization [WHO] Disability Assessment Scale 2.0 [WHODAS 2.0] mean score, 2.3 ± 0.8 vs. 2.4 ± 0.8, p = .55).
Measures
Clinical complexity was measured with the clinician-assessed C4 Inventory, a checklist of 35 characteristics of a person, their illness, and their environmental context (33 items plus 2 write-in “other” characteristics). The psychiatrist completes this checklist after a psychiatric assessment by checking each characteristic which was found to be present to systematically record sources of complexity. The C4 Inventory yields a list of characteristics that are endorsed for a given patient and a summary score (total number of checked characteristics). If a clinician lacks information about an item (e.g., if the question is not asked), the item is not checked. The development of items for the C4 Inventory included a literature review and iterative review by experienced clinicians to identify characteristics that are “both common and associated with difficulty in making a diagnosis, higher than typical intensity or duration of intervention, greater number of interventions, or poorer than typical outcomes of treatment” (Maunder et al., 2016). As such, these items were highly aligned with the definition of complexity quoted above. In a validation study, the intraclass correlation for comparison of summary score results between the two clinicians was .74 (p < .001), the difference between different clinicians’ summary scores was not significant, and the mean item-level agreement between raters was 31.5 (out of 35 items for possible agreement). Convergent validity was indicated by significant correlations between the C4 Inventory summary score and (a) lower Global Assessment of Function (Spearman’s rho = −0.44), (b) number of psychiatric diagnoses, (c) psychiatric distress measured with the K10 inventory (r = .36, p < .001), and (d) overall function measured with WHODAS 2.0 (r = .31, p < .001) (Maunder et al., 2016). For the current analysis, the “other” items were excluded, leaving 33 items for analysis.
Child abuse or neglect and childhood loss were assessed using two items from the C4 Inventory. These items note that the assessing physician identified “any childhood abuse or neglect” or “significant childhood loss (e.g., death of a parent, sibling) or permanent separation of parents.” Because of the checklist format of the C4 Inventory, these items are dichotomous (yes/no).
Psychological distress was measured with the K10, a self-report completed by the patient. The K10 has five items probing anxiety symptoms and five items probing depressive symptoms, each measured on a 5-point scale (1-5), which are summed to create a scale from 10 to 50 (R. C. Kessler et al., 2002). Disability was assessed with the 12-item version of the WHODAS 2.0, a self-report scale completed by patients in which disability in 12 domains is rated from none (0) to extreme (6) (Gold, 2014). Overall disability was calculated as the mean of all items.
Analysis
Demographic (sex, age) and clinical (categories of psychiatric diagnoses) data were tabulated and compared between participants who had either childhood abuse, neglect, or loss or none of these using a chi-square test. The relationship of each of childhood abuse or neglect and childhood loss with other characteristics that contribute to complexity was tested using logistic regression, entering sex, age, and presence or absence of depressive mood disorder, PTSD and other stress-response syndromes, substance use disorder, or personality disorder. These diagnostic categories are expected to be related to childhood adversity based on previous research and were confirmed to be significantly related to childhood adversity in the current sample. Odds ratios and the 95% confidence intervals for odds ratios were computed as a measure of effect size. The Bonferroni correction for 31 comparisons was used, setting the level of significance in this analysis at p (.05 / 31) ≤ .0016. Because complexity items are interrelated, the Bonferroni correction is a conservative standard.
To test the relationship between childhood adversity and overall psychiatric complexity, the total number of C4 characteristic endorsed (excluding the two childhood items) was summed. The relationship of this total complexity score with each item of childhood adversity was assessed by comparing median values in dependent samples (χ2).
The relationship between childhood abuse and psychological distress (K10 score) or disability (WHODAS 2.0 score) was tested using analysis of variance, with age, sex, depressive mood disorder, substance use disorder, PTSD and other stress-response syndromes, and personality disorder entered as covariates.
Results
Table 1 describes the demographic and clinical characteristics of these patients.
Demographic and Clinical Characteristics of Participating Patients.
Childhood abuse or neglect was identified by the assessing psychiatrist in 474 patients (36.0%). Significant childhood loss was present in 236 patients (17.9%). As detailed in Table 1, patients with childhood abuse, neglect, or loss were more likely to be female, older, and diagnosed with mood disorders (excluding bipolar disorder), substance use disorders, stress-response disorders (including PTSD), and personality disorders.
The relationship of aspects of complexity to childhood abuse or neglect and childhood loss is shown in Table 2. Correcting for 31 comparisons (setting significance at p < .0016), childhood abuse or neglect was significantly associated with 11 of the 31 other indices of clinical complexity. The prevalence of these indices in patients with or without childhood abuse or neglect and the odds ratio of the difference is illustrated for the complexity indices in which the difference is significant in Figure 1. Odds ratios for these contributors to complexity range from 1.7 (adult trauma) to 5.0 (lack of local resources for geographic reasons). After correcting for multiple comparisons, childhood loss was not significantly associated with individual indices of clinical complexity (Table 2).
Relationship of Clinically Significant Childhood Abuse and Neglect and of Childhood Loss With Other Indices of Clinical Complexity.
Note. Controlling for age, sex, mood disorder, substance use disorder, PTSD or stressor-related disorder, and personality disorder. OR = odds ratio; CI = confidence interval; PTSD = posttraumatic stress disorder.
Prevalence of indices of complexity in patients with or without childhood abuse or neglect and odds ratio of difference.
With respect to overall complexity (the summed total number of indices of complexity endorsed by the assessing clinician excluding childhood abuse and neglect and childhood loss), childhood abuse or neglect was associated with overall complexity (no childhood abuse or neglect: median 3 indices of complexity, interquartile range 1-5; childhood abuse or neglect: median 5 indices of complexity, interquartile range 3-8; independent samples median test χ2 = 136, p < .001). Childhood loss was also significantly associated with overall complexity (no childhood loss: median 3 indices of complexity, interquartile range 2-6; childhood loss: median 5 indices of complexity, interquartile range 3-8; independent samples median test χ2 = 38, p < .001). The relationship between childhood abuse or neglect, childhood loss, and their interaction with total number of complexity indices endorsed is shown in Figure 2.
Relationship between childhood abuse or neglect, childhood loss, and their interaction on number of complexity indices endorsed.
Analysis of variance indicated that covarying for age, sex, and presence of mood disorder, PTSD, substance disorder and personality disorder, childhood abuse, and neglect was significantly associated with psychological distress (K10 score, F = 6.2, p = .01) and disability (WHODAS 2.0 score, F = 5.0, p = .03). Neither of these measures was significantly associated with childhood loss or with the interaction of childhood abuse or neglect and childhood loss.
Discussion
This study demonstrates that among outpatients referred for assessment to psychiatrists in an academic general hospital, childhood abuse and neglect are common and are associated with a wide range of characteristics of the person, their psychiatric illness, and their environment which together contribute to clinical complexity. As the presence of clinical complexity predicts that standard or guideline-based treatments will result in suboptimal outcomes, this means that screening for childhood adversity is an important clinical task to alert psychiatrists to the likelihood that more intensive, patient-centered, and trauma-informed treatments may be necessary. Importantly, PTSD and related diagnoses were present in a minority of these patients (4.1% of patients with no childhood abuse, neglect, or loss; 20.2% of those with this exposure), indicating that the impact of childhood adversity on complexity goes well beyond the domain of trauma-related diagnoses.
In this cohort of patients, each of childhood abuse or neglect and childhood loss was associated with manifesting a greater number of other characteristics that contribute to complexity and if both childhood neglect or abuse and loss were reported, the number of other indices of complexity was even higher (Figure 2). This is important because it is the nature of clinical complexity that multiple factors interact in unpredictable ways. When assessing adults, appreciating their experiences of childhood adversity may deepen understanding of the sources of their complex circumstances, and may also serve as a marker or flag that indicates that other clinical and sociodemographic factors are likely to be at play, contributing to a complex presentation and demanding treatment resources that go beyond the standards recommended by treatment guidelines.
It is noteworthy that childhood abuse or neglect was associated with 11 individual contributors to complexity, whereas the relationship of childhood loss with these characteristics was not statistically significant. This may indicate that experiences with abuse and neglect were a more potent “cause of causes” in this cohort compared to experiences with loss. Previous studies of the specificity of the impact of different types of childhood adversity have had inconsistent results. For example, in the U.S. National Comorbidity Survey, the relationship between childhood adversity and adult psychopathology was not specific to type of adversity (R. C. Kessler, Davis, & Kendler, 1997). On the contrary, in the Netherlands Study of Depression and Anxiety, childhood trauma rather than other adverse childhood events (including death or divorce of parents) was associated with adult anxiety and depression (Hovens et al., 2010). To our knowledge, no previous studies have compared the relationship of specific types of childhood adversity with indices of medical complexity in adulthood.
The C4 Inventory is comprised of a collection of items that experts considered to be related to complexity, which was subsequently validated empirically through tests of convergent validity (Maunder et al., 2016), and which are not necessarily theoretically related (just as an inventory of surgical mortality might include items for advanced age of the patient and duration of anesthesia, which would be expected to interact in predicting the outcome but not necessarily to be positively correlated). As such, reasons for associations between childhood adversity and some specific items in the current study may be more or less obvious. In some cases, these associations are predicted by previous research, for example, the association to recent harm to self (Afifi et al., 2008), multiple diagnoses (Sansone, Songer, & Miller, 2005), internalizing traits (Heleniak, Jenness, Stoep, McCauley, & McLaughlin, 2016), impulsivity (Daray et al., 2016), adult trauma (Arata, 2000), and aspects of interpersonal isolation and insecure attachment (Murphy et al., 2014). In other cases, plausible paths are apparent. For example, in our setting, childhood adversity is more common among patients with HIV (human immunodeficiency virus), whose lives are very often complicated by discrimination or stigma. Inadequate housing and poverty may both contribute to and result from mental illness (Costello, Compton, Keeler, & Angold, 2003), and similarly be bidirectionally related to other aspects of complexity. The reason for the association of childhood abuse and geographical distance from the clinic/lack of local resources is less apparent. This may relate to a referral bias—that less complex presentations of mental illness are unlikely to require a referral to a distant academic hospital.
One reason that it may be useful to identify aspects of patients’ circumstances that contribute to complexity above and beyond psychiatric diagnoses is that identification may facilitate developing programs and interventions that target these sources of vulnerability specifically, as opposed to interventions organized by diagnosis as often occurs in mental health care. Figure 1 illustrates a contrast between characteristics of complexity that are of low prevalence but strongly related to childhood abuse or neglect (e.g., lack of local services for geographic reasons, experience of discrimination or stigma) and those that are more weakly related, but of much higher prevalence (e.g., traits of lack of affiliation and internalizing, interpersonal isolation). The latter may be better candidates for program development as they are likely to be more widely applicable. Identifying factors that are associated with finding less benefit from standard mental health treatments may lead to the development of interventions that are more patient-centered and less prone to the biases of health care providers, such as those developed “in poor communities . . . specifically to address the contextual stressors in low-income people’s lives” which have produced positive outcomes (Goodman, Pugach, Skolnik, & Smith, 2013).
With respect to the relationship of childhood adversity to aspects of diversity in this group of patients, it is noteworthy that clinically significant adversity was more common among females and older patients. Childhood adversity was also linked to indicators of sociocultural marginalization: experience of significant discrimination or stigma, inadequate housing or poverty, and geographic isolation. Childhood adversity was not associated with language barriers that were judged likely to interfere with assessment or treatment. Descriptions of race, ethnicity, and gender identity are not collected in this database.
One must be cautious in interpreting these results for several reasons. This is a study of patients referred to a hospital-based mental health clinic for assessment. As such, they may not be representative of all psychiatric outpatients and are certainly not representative of the general population. There are also limitations on the confidence that we can have in these results due to the limits of a naturalistic study of a real-world mental health practice. In particular, the sample studied constitutes only 27% of patients for whom standardized measure were used during assessment. There were significant differences between patients for whom the C4 Inventory was or was not completed, although the size of these differences was modest. Assessment of all items on the C4, including childhood abuse or neglect and childhood loss, depends on the quality of the clinical assessment. A previous study of the interrater reliability of two observers of the same interview using the C4 was acceptably high, but it is not known how consistent the results would be from two independent assessments. As multiple items on the C4 are intended to identify circumstances that are likely to contribute to difficulty with assessment or treatment (in the assessing clinician’s judgment), the resulting prevalence probably underestimates the frequency of occurrence of these characteristics.
This study supports routine clinical assessment of childhood adversity in psychiatric outpatients. Furthermore, the results of this study identify several characteristics of complex psychiatric presentations that are associated with past adversity, above and beyond the impact of mental health syndromes in themselves. Knowledge of these variables can aid the clinician in developing a more in-depth, accurate, and individualized understanding of the patient, which in turn will permit interventions better tailored to their unique circumstances. Furthermore, systems-level resources and interventions that focus on transdiagnostic aspects of complex presentations may be of value for the large number of users of mental health service who have experienced childhood abuse or neglect. These include resources that address psychiatric comorbidity, suicidality and self-harm, impulsivity, internalizing traits (i.e., the propensity toward shame, guilt, hopelessness, and low self-esteem), traits that reduce interpersonal affiliation (i.e., attachment avoidance), interpersonal isolation, lack of personal agency, inadequate housing, poverty, discrimination, stigma, and ongoing trauma.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Dr. Maunder’s work is supported in part by the Chair in Health and Behavior at Sinai Health System, funded by Sinai Health System and the University of Toronto, and by the Medical Psychiatry Alliance.
