1. Method

1.1. Sample

The NESARC is a representative sample of civilian, non-institutionalized adults living in the United States (U.S.). Data from the current study were cross-sectional, and came from the second wave of the NESARC collected between 2004 and 2005. This wave of the NESARC collected information from 34,653 adults aged 20 years and older living in households as well as non-institutionalized group dwellings. The NESARC sample was weighted to adjust for nonresponse, the selection of one person per household, and the oversampling of specific groups. Once weighted, the data were adjusted to be representative of the U.S. population based on several sociodemographic indicators (e.g., region, age, sex, race/ethnicity) based on 2000 U.S. Census data (Grant et al., 2004). Data were collected through face-to-face interviews by trained lay interviewers of the U.S. Census Bureau and the response rate was 86.7%. Additional details of the NESARC have been described elsewhere (Grant et al., 2004, Grant et al., 2003, Ruan et al., 2008).

1.2. Measures

1.2.1. Emotional maltreatment

Emotional abuse and emotional neglect were assessed using questions adapted from the Adverse Childhood Experiences (ACE) Study (Dong et al., 2003, Dube et al., 2003). Respondents were asked to report on experiences that occurred before the age of 18 years. The emotional abuse items were measured on a 5-point ordinal scale (never, almost never, sometimes, fairly often, and very often). Emotional abuse was defined as having “fairly often” or “very often” experienced a parent or other adult living in the home: (1) swear at or insult the respondent; (2) threaten to hit or throw something at the respondent, but didn’t do it; or (3) act in any other way that made the respondent feel afraid. The emotional neglect items were assessed on a similar 5-point ordinal scale (never true, rarely true, sometimes true, often true, and very often true). Emotional neglect in childhood was assessed with the following 5 items: (1) the respondent felt there was someone in the family who wanted them to be a success; (2) someone in the respondent’s family made them feel special or important; (3) the respondent’s family was a source of strength or support; (4) the respondent felt part of a close knit family; and (5) someone in the respondent’s family believed in them. Consistent with past research, these items were reverse-coded and summed; scores of 15 or greater were considered indicative of having experienced emotional neglect in childhood (Afifi et al., 2011, Dong et al., 2003, Dube et al., 2003). Participants were categorized into four groups based on their childhood experiences of emotional maltreatment (i.e., no emotional maltreatment, emotional neglect only, emotional abuse only, and both emotional neglect and emotional abuse).

1.2.2. Other childhood maltreatment

Other forms of child maltreatment, including physical neglect, harsh physical punishment, physical abuse, sexual abuse, and exposure to intimate partner violence were also assessed using questions adapted from the ACE study (Dong et al., 2003, Dube et al., 2003). Respondents were asked to report on child maltreatment experiences that occurred before the age of 18 years. All child maltreatment items were measured on a 5-point ordinal scale (never, almost never, sometimes, fairly often, and very often). Dichotomous coding was used to indicate the presence or absence of each additional type of child maltreatment. Physical neglect was defined as present if the respondent ever (i.e., any response other than “never” on the 5-point ordinal scale) experienced any of the following four items: (1) been left alone or unsupervised before age 10; (2) gone without needed things such as clothes, shoes, or school supplies because a parent or other adult living in the home spent the money on themselves; (3) been made to go hungry or did not have regular meals prepared; and/or (4) had a parent or other adult living in the home ignore or fail to get the respondent medical treatment. Harsh physical punishment was assessed with the question, “As a child, how often were you ever pushed, grabbed, shoved, slapped, or hit by your parents or any adult living in your house?” Participants with a response of “sometimes” or greater to this question were considered as having experienced harsh physical punishment in childhood. Physical abuse was defined as present if the respondent reported having ever (i.e., any response other than “never” on the 5-point ordinal scale) being hit so hard that it left marks, bruises, or caused an injury by a parent or other adult living in the home. Sexual abuse was defined as present if the respondent reported having ever (i.e., any response other than “never” on the 5-point ordinal scale) experienced any unwanted sexual touching or fondling, or any attempted or actual intercourse by an adult or other person that was unwanted or occurred when the respondent was too young to understand what was happening. To determine whether the respondent was exposed to intimate partner violence in childhood, respondents were asked whether their mother’s partner had: (1) pushed, grabbed, slapped, or thrown something at their mother, (2) kicked, bit, or hit their mother with a fist or something hard, (3) repeatedly hit their mother for at least a few minutes, or (4) threatened her with a knife or gun, or use a knife or gun to hurt her. Any response of “sometimes” or greater to items 1 or 2, or any response other than “never” to items 3 or 4, was considered indicative of having being exposed to intimate partner violence in childhood. Maternal perpetration of intimate partner violence was not assessed in the survey.

1.2.3. Family history of dysfunction

The family history of dysfunction variable was also based on questions from the ACE study (Dong et al., 2003, Dube et al., 2003). Any respondent reporting that a parent or other adult living in the household had a problem with alcohol or drugs, went to jail or prison, was treated or hospitalized due to mental illness, attempted or completed suicide was coded as having experienced family dysfunction in childhood.

1.2.4. Mental disorders

Lifetime diagnoses of Axis I and Axis II disorders were made using the Alcohol Use Disorder and Associated Disablities Interview Schedule – Diagnostic and Statistical Manual of Mental Disorders (DSM) – Fourth Edition (AUDADIS-IV; American Psychiatric Association, 1994; Grant et al., 2003; Ruan et al., 2008). Reliability and validity of the AUDADIS-IV have been established (Grant et al., 2004, Grant et al., 2003, Ruan et al., 2008). Axis I disorders included: (1) mood disorders (major depression, dysthymia, mania, and hypomania), (2) anxiety disorders (panic disorder with or without agoraphobia, social phobia, specific phobia, generalized anxiety disorder, post-traumatic stress disorder), and (3) substance use disorders (alcohol abuse or dependence, prescription or illicit drug use or dependence). Axis II disorders included: (1) cluster A personality disorders (paranoid, schizoid, schizotypal), (2) cluster B personality disorders (antisocial, borderline, histrionic, narcissistic), and (3) cluster C personality disorders (avoidant, dependent, obsessive–compulsive).

1.2.5. Sociodemographic covariates

Sociodemographic covariates included sex (male/female), age (years), marital status (married/common-law, separated/divorced/widowed, never married/single), race/ethnicity (American Indian/Alaska Native, Black, Hawaiian/Pacific Islander, Hispanic, White), education (less than high school, high school, some college/university, post-secondary degree), and past year household income ($19,999 or less, $20,000 to $39,999, $40,000 to $69,999, $70,000 or more). Sociodemographic covariates were chosen based on their association with childhood emotional maltreatment (Black et al., 2001, Iwaniec et al., 2006) and/or mental disorders (Baumeister and Härter, 2007, Kessler et al., 2005).

1.3. Data analysis

Statistical weights were applied in all analyses to ensure that the NESARC data were representative of the general U.S. population. To account for the complex sampling design of the NESARC, Taylor series linearization was used as a variance estimation technique using SUDAAN software (Shah, Barnwell, & Biehler, 2004). First, descriptive statistics were computed to examine the distribution of sociodemographic characteristics, other types of child maltreatment, family history of dysfunction, and Axis I and Axis II mental disorders by childhood emotional maltreatment status. Second, logistic regression models were computed to examine the association between emotional maltreatment status and Axis I and Axis II mental disorders. These models compared emotional neglect relative to no emotional maltreatment, emotional abuse relative to no emotional maltreatment, both emotional neglect and emotional abuse relative to no emotional maltreatment, and emotional neglect relative to emotional abuse. Models were first adjusted for sociodemographic covariates (AOR-1), and then further adjusted for other forms of child maltreatment and family history of dysfunction (AOR-2). Finally, sex by emotional abuse and sex by emotional neglect interactions were examined in relation to Axis I and Axis II mental disorders. To protect against potential for error associated with multiple comparisons, results at a p-value of less than .01 (and corresponding 99% confidence intervals) were considered statistically significant.

2. Results

Relationships were attenuated after further adjustment for other forms of child maltreatment and family history of dysfunction. However, a substantial proportion of these relationships remained significant in the fully adjusted models. Experiencing childhood emotional neglect only was associated with increased odds of major depression (AOR-2 = 1.3; 99% Confidence interval [CI] = 1.1–1.6), dysthymia (AOR-2 = 1.8; 99% CI = 1.3–2.3), any mood disorder (AOR-2 = 1.2; 99% CI = 1.0–1.5), and social phobia (AOR-2 = 1.4; 99% CI = 1.1–1.7). With the exception of hypomania, experiencing childhood emotional abuse only remained significantly associated with all other individual Axis I disorders assessed in the fully adjusted models (AOR-2 ranged from 1.3 to 2.3). Experiencing both emotional neglect and emotional abuse was associated with increased odds of major depression, dysthymia, mania, any mood disorder, panic disorder, social phobia, generalized anxiety disorder, post-traumatic stress disorder, and any Axis I disorder (AOR-2 ranged from 1.4 to 2.2). A direct comparison of emotional abuse only relative to emotional neglect only indicated that childhood emotional abuse was associated with significantly higher odds of a lifetime diagnoses of several specific Axis I mental disorders (i.e., major depression, mania, specific phobia, post-traumatic stress disorder, and alcohol abuse/dependence) relative to experiencing childhood emotional neglect only (AOR-2 ranged from 1.3 to 1.7).

After further adjustment for other forms of child maltreatment and family history of dysfunction, experiencing childhood emotional neglect only remained significantly associated with increased odds of schizoid, schizotypal, any cluster A, borderline, avoidant, and any Axis II personality disorder (AOR-2 ranged from 1.2 to 1.8). Experiencing childhood emotional abuse only remained associated with significantly increased odds of 9 of 10 individual (i.e., paranoid, schizotypal, antisocial, borderline, histrionic, narcissistic, avoidant, dependent, and obsessive compulsive) personality disorders assessed (AOR-2 ranged from 1.4 to 2.8) and experiencing both emotional neglect and emotional abuse in childhood remained associated with significantly increased odds of 6 of 10 individual (i.e., paranoid, schizoid, antisocial, borderline, avoidant, obsessive–compulsive) personality disorders assessed (AOR-2 ranged from 1.5 to 3.0). A direct comparison of emotional abuse only relative to emotional neglect only indicated that childhood emotional abuse was associated with significantly higher odds of a lifetime diagnoses borderline personality disorder (AOR-2 = 1.6; 99% CI = 1.1–2.4), narcissistic personality disorder (AOR-2 = 1.5; 99% CI = 1.1–2.2), any cluster B disorder (AOR-2 = 1.4; 99% CI = 1.1–1.9), obsessive–compulsive personality disorder (AOR-2 = 1.5; 99% CI = 1.0, 2.1), and any Axis II disorder (AOR-2 = 1.4; 99% CI = 1.1, 1.8) relative to experiencing childhood emotional neglect only.

Finally, all sex by emotional maltreatment status interaction terms were non-significant at p < .01, indicating that childhood experiences of emotional maltreatment have similar effects on both men and women.

3. Discussion

This study contributes to the existing literature on the childhood emotional maltreatment in several ways. First, 14% of the U.S. general population reported experiencing emotional maltreatment from a parent or other caregiver living in the home before the age of 18 years. The most prevalent form of child maltreatment was emotional neglect only (6.2%), followed by emotional abuse only (4.8%). The least prevalent pattern was experiencing both childhood emotional abuse and neglect (3.1%). Second, childhood emotional maltreatment, particularly among participants experiencing both emotional abuse and emotional neglect, exhibited a high rate of co-occurrence with other forms of child maltreatment (ORs ranged from 2.1 to 68.0). Of particular note is the relationship between emotional abuse only and both emotional abuse and neglect with harsh physical punishment (ORs of 22.7 and 49.9, respectively) and physical abuse (ORs of 29.6 and 68.0, respectively). Third, all categories of emotional maltreatment were associated with increased odds of several Axis I and Axis II mental disorders independent of sociodemographic characteristics, other forms of child maltreatment, and a family history of dysfunction. Finally, the effects of childhood emotional abuse appeared to be stronger than the effects of childhood emotional neglect for many outcomes.

The finding that 14% of the U.S. adult population reported experiencing childhood emotional maltreatment is similar to prevalence estimates from other representative samples (Afifi et al., 2011, Finkelhor et al., 2010, Johnson et al., 2000, Waxman et al., 2014). However, across the literature, prevalence estimates vary widely due to differences in sample characteristics, how emotional maltreatment is defined, the specific type of emotional maltreatment assessed, and the operationalization of emotional maltreatment constructs. For example, of child maltreatment cases reported to the National Child Abuse and Neglect Data System (NCANDS) in the U.S., the mean number of child emotional abuse victims was 11.7 per 10,000 children 0 to 18 years old (Hamarman, Pope, & Czaja, 2002). It is significant to note that there was more than a 300-fold variation in rates of emotional abuse among the 43 states reporting emotional abuse to NCANDS (Hamarman et al., 2002). To date, no uniform legal definition of what constitutes emotional maltreatment exists, there is a lack of consensus regarding the definition and measurement of emotionally abusive and neglecting parental actions, and a “gold standard” measure of emotional maltreatment has yet to be developed (Baker, 2009, Glaser, 2011, Hamarman et al., 2002, Tonmyr et al., 2011, Trocmé et al., 2011). Resolving these issues will help integrate findings across studies and advance the body of knowledge on emotional maltreatment (Baker, 2009).

Within the child welfare system, emotional maltreatment tends to be more chronic than other forms of maltreatment substantiated by child protective organizations (Chamberland et al., 2012). However, cases of emotional maltreatment, especially in a single form, are less visible, less prioritized, and less subject to protective service intervention (Chamberland et al., 2011). For example, emotional maltreatment represents the primary type of substantiated maltreatment in only 9% of all cases investigated by child welfare agencies in Canada (Public Health Agency of Canada, 2010). In this study, we found that respondents were slightly more likely to report experiencing emotional neglect than emotional abuse in childhood, and there is evidence suggesting that emotional neglect is less likely to come to the attention of child welfare agencies than emotional abuse. Data from the Canadian Incidence Study of Reported Child Abuse and Neglect indicate that reports of emotional abuse were almost 2.5 times more frequent than reports of emotional neglect across child welfare agencies in Canada (Chamberland et al., 2011). The decreased detection of, and intervention with, cases of emotional neglect could be because acts of omission (e.g., emotional neglect) are more difficult to identify than acts of commission (e.g., emotional abuse) (Chamberland et al., 2011). Recognizing and responding to emotional maltreatment in all its forms remains an important public health priority.

Researchers have speculated that emotional maltreatment is a core component of all forms of child maltreatment (Garbarino et al., 1986, Hart and Brassard, 1987), and the high rate of co-occurrence of emotional maltreatment with other forms of child maltreatment and family dysfunction found in this study lends support to this assertion. While emotional maltreatment can occur in the absence of other maltreatment, the reverse is less common (Chamberland et al., 2011, Gibb et al., 2007b). The specific effects of emotional maltreatment are often confounded by the high association between different types of abuse (Bifulco et al., 2002), and the failure to account for co-occurring forms of child maltreatment in analyses may lead to biased estimates of the association between childhood emotional maltreatment and adult mental health. Emotional maltreatment has also been found to co-occur with other dysfunctional parenting practices (Bifulco et al., 2002, Taillieu and Brownridge, 2013). Given the high correlation between childhood emotional maltreatment, other forms of child abuse, and dysfunctional family environments reported in this study and elsewhere (Bifulco et al., 2002, Bruce et al., 2012, Finza-Dottan and Karu, 2006, Taillieu and Brownridge, 2013, Vachon et al., 2015, Waxman et al., 2014), the presence of emotional maltreatment should alert professionals to the possibility of other forms of maltreatment and dysfunctional parenting occurring in the family.

Disentangling the specific effects of various forms of child maltreatment has important implications for intervention and treatment strategies. If specificity of effects are noted (i.e., specific types of maltreatment are associated with specific disorders) then researchers can focus on identifying potential mediators (Gibb, Chelminski, et al., 2007). However, if effects are non-specific (i.e., specific types of maltreatment are associated with a variety of different disorders), then researchers should focus on moderators, and factors contributing to development of one disorder relative to another become more important (Gibb, Chelminski, et al., 2007). In this study, the relationship between emotional maltreatment and mental disorders appeared to be somewhat contingent on the type of emotional maltreatment (abuse vs. neglect) experienced in childhood.

Childhood emotional neglect was associated with lifetime diagnoses of specific mental disorders (i.e., major depression, dysthymia, social phobia, as well as schizoid, schizotypal, borderline, and avoidant personality disorders). These disorders seem to suggest social withdrawal and avoidance of, or difficulty with, interpersonal relationships. The lack of love, affection, and support that characterizes emotional neglect likely compromises a child’s ability to form a secure attachment with caregivers (Ainsworth, 1979), which is hypothesized to provide the foundation for later individual differences in personality development and psychopathology (Bowlby, 1973). Once an attachment pattern is formed, it tends to endure and forms the basis of attachment patterns through the development of internal representations, or “working models” of self and others throughout the lifespan (Ainsworth, 1979). Therefore, the mechanism linking childhood emotional neglect to specific disorders could be through its impact on early attachment processes. Other potential mediators of the relationship between childhood emotional maltreatment (primarily emotional abuse) and later mental health problems that have received some support in the literature include emotional dysregulation (Coates & Messman-Moore, 2014), immature coping styles and low self-esteem (Finza-Dottan & Karu, 2006), hopelessness (Hamilton et al., 2013), and negative automatic thoughts (Gibb et al., 2007a, van Harmelen et al., 2010).

In contrast, the effects of childhood emotional abuse appeared to manifest across a broad spectrum of mental disorders across the lifespan. This could be evidence for the non-specificity of effects of childhood emotional abuse on adult mental health. While emotional neglect is a more singular construct (e.g., denying emotional responsiveness to the child), emotional abuse encompasses a variety of different parental actions (e.g., threats, belittling, degradation, hostility) and it could be that specific types of emotional abuse have specific effects on mental health. For example, terroristic types of behavior likely generate fear and threats to physical safety whereas spurning might have more of an impact on self-esteem and self-acceptance (Chamberland et al., 2012). Future research examining specific subtypes of emotional abuse may help to clarify the relationship between childhood emotional abuse and adult mental health.

This study also found that childhood experiences of emotional abuse appear to be particularly damaging to long-term mental health, and emotional abuse was more strongly associated with many of the mental disorders in this study than emotional neglect. Emotional abuse may be particularly harmful to long-term mental health because negative cognitions are directly given to child (e.g., you are worthless), whereas for other forms of maltreatment, including emotional neglect, the child has to find their own attributions regarding the cause (Gibb, 2002). That is not to say that emotional neglect is not harmful; it was clearly associated with poor mental health outcomes in this study. It could also be that emotional neglect is more strongly associated with more externalizing problem behaviors (e.g., criminality, aggression, impulsivity) that were not assessed in this study. There is evidence suggesting that parental rejection and low parental warmth and responsiveness are associated with externalizing problem behaviors among children and adolescents (Muris et al., 2003, Roelofs et al., 2006, White and Renk, 2012), and it could be that childhood emotional neglect manifests as externalizing problem throughout the lifespan.

Strengths of this study include the inclusion and comparison of both active (i.e., emotional abuse) and passive (i.e., emotional neglect) forms of emotional maltreatment, the assessment of a broad range of mental disorders, and the use of a nationally representative dataset. This study is also subject to a number of limitations that should born in mind when extrapolating from results. First, the data were cross-sectional, which precludes establishment of the temporal sequence of constructs and makes inferences about causality impossible. Second, the retrospective nature of the data collection introduces the possibility of recall bias. It is important to note that participants were asked to report on childhood experiences that may have occurred decades earlier, and the time interval between when child maltreatment occurred and the data were collected varied substantially across participants. In addition, current mental health status may have influenced perceptions of childhood experiences. However, there is evidence supporting the validity of accurate recall of adverse childhood events (Hardt & Rutter, 2004). Third, although mental disorder diagnoses were made by a reliable structured interview conducted by trained lay interviews, this assessment approach may not match the accuracy of an experienced clinician. Fourth, the assessment of emotional abuse and emotional neglect were limited to variables available in the dataset. It is widely accepted that five main types of emotional maltreatment exist: spurning, terrorizing, corrupting/exploiting, isolating, and denying emotional responsiveness (American Professional Society on the Abuse of Children, 1995, Hart and Brassard, 1987). Although we were able to tap into (albeit limited) aspects of spurning, terrorizing, and denying emotional responsiveness in this study, measures of corrupting/exploiting and isolating components of emotional maltreatment were not assessed in the dataset. More comprehensive assessments of all components of emotional maltreatment are an important avenue for future research. As well, the failure to collect information of intimate partner violence perpetrated by female caregivers represents an important source of bias in this study. Finally, proposed mechanisms linking childhood emotional maltreatment to adult mental disorders were not considered in this study. Longitudinal data are required to better understand the influence of early childhood experiences on subsequent functioning and behavior.

Given the high prevalence of childhood emotional maltreatment reported in this study and its association with other forms of child maltreatment and compromised adult mental health, parents and other caregivers need to be made aware of the harmful effects that emotional maltreatment has on long-term mental health. However, a review of universal evidenced-based parenting programs found that most contained no content specific to emotional maltreatment (Baker, Brassard, Schneiderman, Donnelly, & Bahl, 2011). The only type of emotional maltreatment that was consistently addressed in all of the programs reviewed was spurning/belittling. This is concerning given a growing body of research linking childhood experiences of emotional maltreatment to a variety of adverse outcomes. A focus on the reduction of non-physical forms of parental aggression needs to be integrated into both efforts to promote positive parenting and to reduce child maltreatment.

4. Competing interests

None to declare.