Is Emotional Lability Distinct From “Angry/Irritable Mood,” “Negative Affect,” or Other Subdimensions of Oppositional Defiant Disorder in Children With ADHD?

Abstract

Objective: Emotional lability (EL) and oppositional defiant disorder (ODD) frequently co-occur with ADHD. This study evaluates whether EL merely represents the negative “mood/affect” component of ODD or forms a distinct dimension. Method: EL and ODD data from 1,317 ADHD participants were analyzed using exploratory factor analysis (EFA) and confirmatory factor analysis (CFA) for binary data. Results: Within ADHD, 39.4% children had ODD and 42.6% had EL. A sizable proportion expressed only either ODD or EL: 16.6% had ODD-only, 19.7% had EL-only, and 22.9% expressed both. In both EFA and CFA, EL forms a separate dimension from ODD items and the “mood/affect” subdimensions (whether classified by Diagnostic and Statistical Manual of Mental Disorders [5th ed.; DSM-5] or the Burke et al. models or the de novo ODD subdimensions derived from our data). This factorial structure remains invariant across gender. Conclusion: EL is distinct from ODD and its “mood/affect” subdimensions. In line with emerging evidence, our findings provide further evidence of factorial validity for EL as a separate construct from ODD.

Keywords

ADHD emotional lability ODD factor analyses

Introduction

ADHD is a common neurodevelopmental disorder characterized by developmentally inappropriate levels of inattention, hyperactivity, and impulsivity. ADHD frequently co-occurs with severe emotional lability (EL; such as frustration intolerance, mood swings, temper outbursts, and emotional fragility) and oppositional defiant disorder (ODD; such as negative mood/affect, conflicts, and oppositional behaviors). Among ADHD, the prevalence of EL is about 30% (Sobanski et al., 2010), whereas ODD is around 30% to 80% (Yüce, Zoroglu, Ceylan, Kandemir, & Karabekiroglu, 2013). Given that a significant correlation exists between ODD and EL symptoms among ADHD (Sobanski et al., 2010), their co-occurrence could arise from their construct overlaps; for example, some ODD items¹ also describe emotional dysregulation behaviors, suggestive of item similarity on face validity. However, it remains unclear whether ODD is a conglomeration of distinct subprocesses, which give rise separately to emotional dysregulation and behavioral dysfunctions, or whether ODD captures a unifying substrate underlying global dysregulation difficulties (Burke, Hipwell, & Loeber, 2010). There is a possibility that severe EL merely represents the negative “mood/affect” component of ODD, if the latter could be accurately cleaved from its behavioral components.

The multidimensionality of the ODD construct has been demonstrated by recent research, showing factorial validity, concurrent validity, and predictive validity of separate and distinct subdimensions (Burke et al., 2010; Stringaris & Goodman, 2009). Indeed, Diagnostic and Statistical Manual of Mental Disorders (5th ed.; DSM-5; American Psychiatric Association [APA], 2013) classifies ODD items under three distinct categories: Angry/Irritable Mood, Argumentative/Defiant Behavior, and Vindictiveness (APA, 2013). Since the publication of the DSM-5, further evidence has been provided on the factorial validity of the DSM-5 grouping (Krieger et al., 2013). Furthermore, a recent study has confirmed the concurrent validity of these three distinct categories demonstrating that Angry/Irritable Mood was associated with emotional problems, Argumentative/Defiant Behavior with ADHD symptoms, and Vindictiveness with conduct problems (Mandy, Roughan, & Skuse, 2014). However, the precise methods of cleaving ODD items into subdimensions are still subjected to debate. Subdimensions of ODD items are defined differently by two key methods. Stringaris and Goodman (2009) proposed a grouping, which is the same as that adapted by the DSM-5. In this first method, “loses temper,” “angry,” and “touchy” items are grouped as Irritable (equal to Angry/Irritable Mood in DSM-5); “argues,” “defies,” “blames,” and “annoys” items are grouped as Headstrong (equal to Argumentative/Defiant Behavior in DSM-5); whereas “spiteful” and “vindictive” items are grouped as Hurtful (equal to Vindictiveness in DSM-5). These reported subdimensions were classified based on predictive validity: The Irritable dimension was found to predict later emotional disorders (depression and anxiety), whereas Headstrong predicts ADHD and Headstrong and Hurtful together predict conduct disorder (Stringaris & Goodman, 2009). In contrast, in the second method proposed by Burke et al. (2010), “touchy,” “angry,” and “spiteful” items are grouped as Negative Affect; “loses temper,” “argues,” and “defies” items as Oppositional Behavior; whereas “annoys” and “blames” items as Antagonistic Behavior. These subdimensions were first extracted by exploratory factor analysis (EFA), and then validated by their prediction of differential outcomes, with Negative Affect predicting depression whereas Oppositional and Antagonistic behaviors predicting conduct disorder.

Some research findings, however, suggest that EL and ODD, despite their co-occurrence, are not identical. Furthermore, EL has been suggested to represent an additional core dimension of ADHD (Reimherr et al., 2010; Skirrow & Asherson, 2013). Among ADHD, ODD is present in 46% of those with low EL, 65% of those with moderate EL, and in 79% of those with high EL (Sobanski et al., 2010). In other words, substantial proportions of the sample express only one of these two conditions, despite their significant correlation. Whether EL represents a separate dimension distinct from ODD in the context of ADHD, to our knowledge, has not been formally evaluated by factor analysis in the literature.

Another important consideration in investigating the issue of construct duplication is the emerging importance occupied by emotional dysregulation in ADHD research, where mood- and emotion-related terms such as emotional dysregulation, emotional impulsivity, affective lability, and mood instability have been used interchangeably (Sobanski et al., 2010). Barkley and Fischer (2010) highlighted that emotionally impulsive symptoms are the most impairing aspects of ADHD that persisted into adulthood, contributing to impairments in family and peer relationships as well as financial and driving ability, and consequently proposed emotional dyscontrol as the third dimension of ADHD phenotype. Seymour, Chronis-Tuscano, Iwamoto, Kurdziel, and Macpherson (2014) demonstrated emotion regulation and its impairment mediate the relationship between baseline ADHD and subsequent depression in a community sample of children aged between 9 and 12 years. Childhood ADHD increased the risk of recurring depression 12-fold (Chronis-Tuscano et al., 2010), whereas youths with comorbid ADHD and mood disorders are 3 times more likely to complete suicide (James, Lai, & Dahl, 2004). These recent findings underscore the clinical importance and relevance of emotional dysregulation in ADHD research.

Moreover, on the etiological level, familial and additive genetic substrates accounting for a phenotypic association between ADHD and emotional dyscontrol have also been identified. Surman et al. (2011) identified familiality in “emotional dysregulation-ADHD” phenotype as a “breeding true” familial subtype. Furthermore, Merwood et al. (2014), using a twin study design, decomposed the genetic architecture of “EL” within the context of ADHD symptoms and identified common genetic substrates between EL and ADHD phenotypes. EL was found to share common genetic etiology with core ADHD dimensions, while showing a stronger link between EL and hyperactive/impulsive symptoms; however, the potential construct overlap between EL and ODD could not be ruled out.

Given that EL has been found to be a predictor of depression and poor prognosis and “mood/affect” subdimensions of ODD may also be a predictor of later depression, these two dimensions could be closely linked. The present study sought to examine the precise nature of the relationship between EL and ODD subdimensions within the context of ADHD. This study utilized a clinical sample of Chinese Han (i.e., the ethnic majority group in China to minimize sample heterogeneity) and set out to answer a series of research questions:

Research Question 1: What is the pattern of co-occurrence between EL and ODD among participants with ADHD?

Research Question 2: Is EL distinguishable from the Angry/Irritable Mood subdimension, as classified by the DSM-5 method, or is EL distinguishable from the Negative Affect subdimension, as classified by the Burke et al. method?

Research Question 3: If not, how is EL distinct from other groupings of ODD items?

Materials and Methods

Participants

The clinical sample consisted of 1,338 children and adolescents with ADHD recruited from Peking University Institute of Mental Health. ADHD caseness was classified using the Clinical Diagnostic Interview Scale (CDIS; Barkley, 1998) based on the Diagnostic and Statistical Manual of Mental Disorders (4th ed.; DSM-IV; APA, 1994) criteria. The inclusion criteria were as follows: (a) 6 to 16 years old, (b) full-scale estimated IQ using Chinese Wechsler Intelligence Scale for Children (C-WISC; Gong & Cai, 1993) is greater than 70; (c) medication-naïve, and (d) Chinese Han descent. The exclusion criteria included major neurological disorders, a diagnosis of schizophrenia, pervasive developmental disorder, epilepsy, mental retardation, or other brain disorders. In addition, children with emotion-related disorders diagnosed using CDIS, such as special phobia, social phobia, separation anxiety disorder, generalized anxiety disorder, dysthymia, depression, and bipolar disorder, were also excluded.

The current study was approved by the Ethics Committee of Peking University Institute of Mental Health. Written informed consent was obtained from parents of all participants.

Diagnoses and Assessment

ADHD caseness

CDIS

The Mandarin-Chinese version of CDIS was developed by Peking University Institute of Mental Health (Yang, Wang, Qian, Biederman, & Faraone, 2004). The semistructured interview was conducted with ADHD probands and their parents by trained psychiatrists, yielding three diagnostic categories: inattentive type (ADHD-I), hyperactive–impulsive type (ADHD-HI), and combined type (ADHD-C). The parent who acted as the main carer and with the best knowledge of the child was used as the primary informant. Comorbidities including ODD, tic disorder (TD), and learning problems were also evaluated.

EL

The EL construct was captured by the items derived from the “Emotional Lability” subscale of Conners’ Parent Rating Scale (Goyette, Conners, & Ulrich, 1978; Xu, 1999). The items were rated directly on a 4-point Likert-type scale (0 = never, 1 = sometimes, 2 = often, and 3 = always) by the same parent who was interviewed. The items included were as follows: (a) “mood changes quickly and drastically” (mood swings), (b) “demands must be met immediately—easily frustrated in effort” (frustration), (c) “cries often and easily” (cry), and (d) “temper outbursts, explosive and unpredictable behavior” (explosive). An EL score was calculated by summing across the four items.

For the categorical comparison of co-occurrence between EL and ODD status among ADHD participants, a cutoff threshold of EL score 6 or more was used to yield EL caseness. As there is no published threshold for clinically significant EL in the Chinese population, the “EL” threshold was empirically derived from a community sample, capturing the top 2% in EL scores. A total of 745 typically developed controls (TDCs) were recruited from elementary schools in Beijing. All participants were of Chinese Han descent. Children with more than three ADHD symptoms evaluated by the ADHD Rating Scale–IV (ADHD RS-IV; Su et al., 2006) were excluded because of the potential presence of an ADHD diagnosis. Exclusion criteria included mania, bipolar disorder, other major psychiatric disorders, family history of psychosis, severe physical diseases, and substance abuse (for more details, see Guan et al., 2009). There was no significant difference in age between the TDC and our ADHD sample—(116 ± 20) versus (118 ± 29), Z = −0.74, p = .462—but there was a higher percentage of males in the ADHD sample—85.6% versus 53.4%, χ² = 255.37, p < .001. Setting the cutoff threshold at 6 (i.e., a score of 6 or above) captured the top 2% of individuals, representing a clinically meaningful threshold of severe EL, that is, 2 SDs above the mean.

ODD

The eight DSM-IV ODD symptoms were derived from parents’ report of the child’s symptoms and were captured using the CDIS by interviewers who recorded the same parent’s description of ODD items, as “present” or “absent.” Eight items were evaluated by the same parent about the presence of each ODD symptoms over the course of at least 6 months, and rated as “0 = no (absent)” and “1 = yes (present).” If four or more symptoms were rated as present and lead to significant functional impairments, ODD diagnosis was made.

The ODD items were first grouped into subdimensions by the DSM-5 method and then grouped by the “Burke et al.” method.

In the DSM-5 method (APA, 2013), “loses temper,” “angry,” and “touchy” items are grouped as an Angry/Irritable Mood subdimension; “argues,” “defies,” “blames,” and “annoys” items as an Argumentative/Defiant Behavior subdimension; whereas the “spiteful” item reflected a Vindictiveness subdimension.

In the Burke et al. (2010) method, “touchy,” “angry,” and “spiteful” items are grouped as Negative Affect; “loses temper,” “argues,” and “defies” items as Oppositional Behavior; whereas “annoys” and “blames” items as Antagonistic Behavior.

Statistics

Comparison of demographic and clinical characteristics between excluded and included participants was conducted using either Pearson’s chi-square test or independent-samples t test as appropriate. Co-occurrence of ODD and EL in ADHD participants was computed by cross-tabulation and chi-square statistics.

For de novo EFA and CFA (confirmatory factor analysis), the initial sample was randomly divided into two equal split-half subsamples: the first “learning” sample for EFA and the second “testing” sample for CFA. The data set was split into “learning” and “testing” samples by a random number generator, providing equal distribution of potential measured and unmeasured covariates.

Examining the “learning” sample using EFA allows an empirically derived factor structure to emerge. Examining the “testing” sample using CFA allows validation of that derived factor model. CFA further permits comparison of the emerged model with other existing competing models already reported in the literature (Reis & Judd, 2000). For our present study, EFA for categorical data, via the weighted least squares means and variance adjusted (WLSMV) (Brown, 2006; B. Muthén, du Toit, & Spisic, 1997), was applied in all 12 items (EL and ODD) using the “learning” sample. The second “testing” data set was used to test potential factor structures, including the one emerging from EFA and the other existing competing models already reported in the literature via CFA. In addition, we also test the modified models in which EL items were grouped with the “mood/affect” subdimensions of ODD while keeping all other initial features unchanged (details of these models are described with the results).

Both measures of absolute and relative fit were assessed (Hoelter, 1983). Although Mplus modeling can handle variables of mixed type (i.e., binary and ordinal), we took caution to dichotomize EL items into a binary variable (0/1 = 0, 2/3 = 1) to minimize the scalar effects (i.e., four-level measure of EL items and two-level measure of ODD items) influencing the final EFA and CFA models. All parameters in the model were estimated unless constrained for a priori reasons. Factor analysis for categorical data was implemented (also known as item factor analysis [IFA] or multidimensional item response theory [IRT]) in Mplus, which reports on the goodness of fit of both EFA and CFA models tested. We also reran the models using the original four-level EL items to cross-check the robustness of the binary results. Due to the sensibility of the chi-square to the sample size, the relative chi-square (χ²/df; Hoelter, 1983) of each model is reported along with the root mean square error of approximation (RMSEA; Browne & Cudeck, 1993), the Taylor–Lewis Index (TLI; Bentler & Bonett, 1980), and the Comparative Fit Index (CFI; Bentler, 1990). Should a novel factor structure emerge, a test for gender invariance would be applied. We conducted multiple group CFAs for gender invariance in novel factorial structures. All data analyses were conducted using the SPSS (Version 17.0; SPSS, Inc., Chicago, IL, USA) and Mplus (L. K. Muthén & Muthén, 1998-2011) statistical packages.

Results

Characteristics of the Sample

Of 1,338 participants, 21 were excluded due to missing data. The analyzed sample consisted of 1,317 participants; there were no significant differences between the excluded and analyzed samples in age (t = 1.42, p = .155), gender (χ² = 0.00, p ≥ .999), or ADHD subtypes (χ² = 3.59, p = .141).

Of 1,317 ADHD participants analyzed, 1,125 (85.6%) were male. The mean age was 9.79 years (SD = 2.42). Six hundred sixty-eight (50.7%) children were classified as having the combined subtype of ADHD, 59 (4.5%) having the hyperactive–impulsive subtype, 590 (44.8%) having the inattentive subtype.

Research Question 1: What is the pattern of co-occurrence between EL and ODD among participants with ADHD?

Within the 1,317 participants with ADHD, 39.4% (519) had ODD and 42.6% (561) had EL (dichotomized using the threshold of 6). A sizable proportion expressed only either ODD or EL: 16.6% had ODD-only, 19.7% had EL-only, 22.9% expressed both, and 40.9% had neither (χ² = 83.1, p < .001).

Research Question 3: If not, how is EL distinct from other groupings of ODD items?

To address these questions, both EFA and CFA were conducted using all 12 binary items (four EL and eight ODD). The initial sample of 1,317 participants was randomly split into two halves: 656 in the “learning” sample for EFA and 661 in the “testing” sample for CFA.

For the EFA solution, the Geomin oblique rotation was used (Browne, 2001; Yates, 1987). The one-factor model did not provide adequate fit (relative χ² = 6.56, CFI = 0.92, TLI = 0.90, RMSEA = 0.099). The fit of the two-factor solution was improved (relative χ² = 3.5, CFI = 0.97, TLI = 0.95, RMSEA = 0.062), but it was the three-factor solution that was indicated by the scree plot (Figure 1A) and provided adequate fit (χ² = 1.92, CFI = 0.99, TLI = 0.98, RMSEA = 0.037). The rotated factor loadings are presented in Table 1. Under the three-factor solution, the EL items form a distinct dimension from the ODD items. The latter form two more factors, partially replicating Burke’s structure of negative affect, antagonistic and oppositional behaviors.

Figure 1.

The scree plot of EFA and path diagram of CFA three-factor model (model 8). (A) Scree plot of EFA on the first split-half “learning” sample. (B) Path diagram of the CFA three-factor model (Model 8) derived from the second split-half “testing” sample. Rectangles represent observed items and ellipses represent the latent variables (factors). The parameters printed by the black straight arrows represent the standardized loadings and the coefficients on the curved gray arrows represent the correlations between factors (p < .05 in all cases).

Table 1.

EFA Three-Factor Solution (First Half of the Data).

	Factor loadings
	1	2	3
EL
Cry	0.47	−0.25	0.03
Explosive	0.54	0.04	0.13
Mood swings	0.95	0.00	−0.12
Frustration	0.56	−0.20	0.01
ODD
Loses temper	0.09	0.66	0.22
Argues	−0.02	0.93	−0.01
Defies	−0.01	0.88	−0.02
Annoys	−0.11	0.76	0.08
Blames	0.07	0.30	0.26
Spiteful	0.03	0.25	0.46
Touchy	0.00	−0.21	0.92
Angry	−0.18	0.01	1.06
EFA via weighted least squares with Geomin rotation

Note. Factor loadings in bold indicate comparatively higher value(s). EL symptoms are derived from Conners’ Parent Rating Scale. ODD symptoms are evaluated from the interviewed parents’ report of the child’s symptoms. EFA = exploratory factor analysis; EL = emotional lability; ODD = oppositional defiance disorder.

The next step in our analysis was to verify the three-factor solution, suggested by the EFA, using the second half of the data. For completeness, several other models were tested and the goodness-of-fit indices are presented in Table 2. In particular, Model 1 corresponds to the one-factor model, in which all 12 binary items load on a single factor. Model 2 consists of two dimensions, one for the ODD items and one for the EL items. Models 3 and 4 correspond to the DSM-5 method; in the former, the EL items belong to a fourth dimension, and in the latter, the EL items have been grouped with the Angry/Irritable dimension while keeping all other initial features unchanged. In the DSM-5 model, there was only one indicator (i.e., “spiteful” item) representing the Vindictiveness dimension, and therefore, Vindictiveness was represented as an observed variable rather than as a latent factor. Model 5 corresponds to the Burke et al. method augmented by a fourth dimension for the EL items, whereas in Model 6, the EL items have been grouped with the Negative Affect dimension of the Burke et al. model. Finally, Model 7 is derived from the EFA best fitted model, and Model 8 is a modification of Model 7 where the “spiteful” item is assigned to the Oppositional/Provocative (O/P) rather than Negativity (N) dimension. Model 8 was tested as “spiteful” cross-loaded on the O/P dimension as suggested by the modification indices in Mplus.

Table 2.

Comparison of Factorial Structures of Tested Models With Goodness-of-Fit Indices According to CFA (Second Half of the Data).

	Fitted model
	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6	Model 7	Model 8
	One-factor model	Two-factor model	Initial Stringaris and Goodman (DSM-5) four-factor model	Modified Stringaris and Goodman (DSM-5) EL = A/Ithree-factor model	Initial Burke et al.four-factor model	Modified Burke et al. EL = NA three-factor model	Model indicated by EFA from Beijing data	Best fitted model: CFA from Beijing data (gender invariant, three-factor model)
Spiteful	GF	ODD	V	V	NA	NA	N	O/P
Angry	GF	ODD	A/I	A/I	NA	NA	N	N
Touchy	GF	ODD	A/I	A/I	NA	NA	N	N
Loses temper	GF	ODD	A/I	A/I	O	O	O/P	O/P
Argues	GF	ODD	A/D	A/D	O	O	O/P	O/P
Defies	GF	ODD	A/D	A/D	O	O	O/P	O/P
Annoys	GF	ODD	A/D	A/D	A	A	O/P	O/P
Blames	GF	ODD	A/D	A/D	A	A	O/P	O/P
Cry	GF	EL	EL	A/I	EL	NA	EL	EL
Explosive	GF	EL	EL	A/I	EL	NA	EL	EL
Mood swings	GF	EL	EL	A/I	EL	NA	EL	EL
Frustration	GF	EL	EL	A/I	EL	NA	EL	EL
Goodness of fit
Relative χ²	5.45	3.12	2.99	5.09	2.98	4.89	2.84	2.76
CFI	0.92	0.96	0.97	0.93	0.97	0.93	0.97	0.97
TLI	0.90	0.95	0.96	0.91	0.96	0.92	0.96	0.96
RMSEA	0.082	0.057	0.055	0.079	0.055	0.077	0.053	0.052

Note. CFA = confirmatory factor analysis; DSM-5 = Diagnostic and Statistical Manual of Mental Disorders (5th ed.); EL = emotional lability; A/I = Angry/Irritable Mood; NA = Negative Affect; EFA = exploratory factor analysis; GF = global factor; ODD = oppositional defiance disorder; V = Vindictiveness; N = Negativity; O/P = Oppositional/Provocative; O = Oppositional Behavior; A/D = Argumentative/Defiant Behavior; A = Antagonistic Behavior; CFI = Comparative Fit Index; TLI = Taylor–Lewis Index; RMSEA = root mean square error of approximation.

As presented in Table 2, the goodness-of-fit indices deteriorated when EL was merged with the respective putative “mood/affect” components in both the DSM-5 (Model 3 vs. Model 4) and Burke et al. model (Model 5 vs. Model 6), mirroring the magnitude of change between Model 2 versus Model 1. That is, the fit was not adequate when the EL items were grouped along with ODD items (Models 1, 4, and 6); the fit improved substantially when the EL items were allowed to form a separate dimension.

The best fitted model was Model 8, which supported EL as a separate construct, distinct from N and O/P dimensions. Because Model 8 was the best fitting one, we proceeded with multiple group CFA (B. Muthén & Christoffersson, 1981) with respect to gender. Analysis indicated measurement invariance in thresholds and loadings for boys and girls (χ² of nested comparison = 7.398, df = 6, p = .286). Therefore, the EL and ODD items form separate dimensions for both girls and boys in a similar manner. The corresponding loadings and intercorrelations between the three factors are reported in the path diagram (Figure 1B). The two ODD dimensions correlated strongly (r = .82); in contrast, EL showed lower correlations with ODD dimensions (.55-.62). The factor loadings derived from CFA were broadly consistent with those from the respective EFA, which allowed for factor cross-loadings.

The analysis was repeated by using the EL items in a 4-point Likert-type scale. The model that was suggested by EFA was in fact identical to Model 8. Regarding the CFA results, the fit indices of all models were almost identical to the ones derived when using the binary items (complete results available upon request). These results indicate that Model 8 best describes the factor structure of the 12 items, regardless of the type of data (binary or ordinal) used for the EL items.

Discussion

There are four key findings of this study. First, EL and ODD are common comorbidities in children with ADHD: 39.4% had ODD and 42.6% had EL. Yet, sizable proportions of participants with ADHD only express either the ODD or EL phenotype alone, and only a quarter express both. Second, EL is a separate construct from ODD and is distinct from the “mood/affect” subdimensions of ODD, whether classified by the DSM-5 (i.e., Angry/Irritable Mood), the Burke et al. models (i.e., Negative Affect), or by the model derived from our data (i.e., Negativity). Third, a novel factorial structure of ODD is identified when juxtaposed against EL items. In our data, ODD is best represented by two dimensions, which are designated as Negativity (i.e., “angry” and “touchy” items) and Oppositional/Provocative (i.e., the remaining six items); and this grouping is similar to that of the Burke et al. model. Fourth, the detected three-factor structure (i.e., EL, Negativity, Oppositional/Provocative) was invariant across gender.

As far as the authors are aware, this is the first study that provides factorial validity of EL as a distinct and separate construct from ODD and its subdimensions.

The construct of “emotional lability” examined here is derived from a subscale within the well-validated Conners’ Rating Scale (Conners, 1973; Parker, Sitarenios, & Conners, 1996). It refers to a subset of symptoms denoting emotional instability, frustration intolerance, rapid and drastic changes in mood, emotional fragility of crying easily as well as unpredictable and explosive temper outbursts. Its distinct factorial validity has previously been demonstrated in different samples only in relation to ADHD symptoms (Conners, 1973; Epstein, Cullinan, & Gadow, 1986; Furlong & Fortman, 1984; Margalit, 1983; Merwood et al., 2014; Parker et al., 1996).

Apart from the factorial validity of the EL construct, there are other findings, indicative of the genetic, concurrent, and predictive validity of this construct.

For genetic validity, a family study provides evidence for co-segregation of emotionally dysregulated ADHD phenotype as a distinctive familial subtype (Surman et al., 2011). A quantitative genetic analysis isolates substantial genetic heritability for EL in the context of an ADHD phenotype (Merwood et al., 2014).

For concurrent validity, a recent functional neuroimaging study identified two independent circuitries for executive attentional deficits and EL in participants with ADHD (Posner et al., 2013). Another study focused specifically on EL symptoms and identified abnormal amygdala-cortical functional connectivity in children with ADHD associated with high EL, and the association between EL and the neurocircuitry signature remains unchanged while accounting for ODD status (Hulvershorn et al., 2014). More specifically, high levels of EL were associated with increased positive functional connectivity between bilateral amygdala and medial prefrontal regions and less positive functional connectivity between amygdala and bilateral insula and superior temporal gyrus. The authors suggest that this pattern may represent a specific disruption in emotional control networks in a subset of children with ADHD and high EL, a process independent of ODD symptoms. For functional correlates, the risks of emotional dysregulation are highlighted in that Emotional Impulsiveness symptoms are the most impairing aspects of ADHD, contributing to family, peer relationships, and financial and driving impairments in adult outcomes (Barkley & Fischer, 2010). Their findings are broadly replicated using an analogous construct—Deficient Emotional Self-Regulation (DESR)—characterized by low frustration tolerance, temper outbursts, emotional impulsivity, and mood lability (Surman et al., 2013).

For predictive validity, childhood DESR was found to adversely impact the quality of life in ADHD adults and was associated with significant functional impairments, reduced marital status, and higher risk for traffic accidents and arrests. In a longitudinal study, baseline DESR was found to predict both DESR and persistence of ADHD at 4-year follow-up; and participants with ADHD + DESR had increased risks of ODD, social problems, and psychiatric comorbidities (Biederman et al., 2012). Our findings showing EL as a distinctive dimension are therefore in line with the emerging evidence, which implicates EL symptoms embodying a dimension with specific concurrent and predictive validity.

From our data, a novel factorial structure of ODD was identified when ODD items were juxtaposed against EL items: ODD is best represented by two dimensions, which were designated as Negativity (i.e., “angry” and “touchy” items) and Oppositional/Provocative (i.e., the remaining six items). This grouping is largely similar to that of the Burke et al. model. However, the detected difference may arise from our sample consisting of ADHD participants while the study sample from Burke et al. consists of female community participants. The current study aimed to test for the independence of EL and ODD items and it is worth noting that EL remains a separate construct from ODD items regardless of whether these items were grouped according to the DSM-5, Burke et al., or our de novo models. Our findings can therefore stimulate further research and replication studies on EL.

Our results should be interpreted in the light of several limitations. First, our clinical ADHD sample may not be generalized to children without ADHD. Second, the effect of anxiety, depression, and autism spectrum disorder symptoms in modifying the expression of EL was not evaluated, as these measures were not available for analyses. Our sample also excluded childhood bipolar disorder and was not phenotyped for severe mood dysregulation (SMD). Indeed, an overlap between EL in 10-item Conners (also known as Conners’ Abbreviated Parent Questionnaire) and Prepubertal and Early Adolescent Bipolar Disorder phenotyped by Washington University in St. Louis Kiddie Schedule for Affective Disorders and Schizophrenia (WASH-U-KSADS) was reported (Tillman & Geller, 2005). The extent to which these comorbidities may influence our findings remains untested and therefore unknown. Future studies with samples enriched and phenotyped for these conditions could test for the invariance and generalizability of our findings. Third, our measures of EL and ODD were also derived from the same source, based on parental report; however, they were recorded differently. EL items were rated directly by parents on a paper instrument. ODD items were reported by parents but recorded by interviewers in CDIS. It is unlikely that the detected independence of EL is entirely attributable to the effects of direct and indirect recording of the same source information based on parental reports. However, a replication study using items derived from the same instrument by the same rater could overcome this limitation. Fourth, only one single indicator (i.e., “spiteful” item) was available to represent the Vindictive dimension. Finally, there is no normative data on EL distribution in the Chinese population stratified by age and gender; we therefore relied on the cutoff threshold derived from the TDC sample as a single group. The TDC consists of children with few symptoms, which may potentially yield a lower cutoff threshold, thereby inflating the co-occurrence between EL and ODD. This could potentially bias our findings toward the null hypothesis leading to an inflated overlap between EL and ODD as well as less children with “neither” (i.e., more children classified as expressing severe “EL” by a lower threshold). Yet, our finding was the contrary. This effect, if it exists, would potentially strengthen rather than weaken our finding. Replication studies with European or U.S. population samples using standardized EL T-scores and cutoff threshold will overcome this limitation. Our CFA analyses model all measures as indicators without assigning arbitrary cutoff thresholds; the results likewise do not support EL merging with the Angry/Irritable Mood subdimension or the Negative Affect subdimension.

In conclusion, our findings provide preliminary evidence that EL is distinct from ODD and the “mood/affect” subdimensions of ODD, that is, Angry/Irritable Mood or Negative Affect. In both EFA and CFA, EL formed a separate dimension from ODD items and this remained invariant across gender. In line with other emerging evidence suggestive of EL as an independent construct with specific concurrent correlates and predictive validity, our findings provide further factorial validity evidence for EL as a separate construct from ODD. We therefore recommend more research focusing on EL, as it may represent a separate clinical entity with important diagnostic and management implications.

Footnotes

Acknowledgements

We thank all the participants who participated in this study.

Authors’ Note

Lu Liu and Wai Chen contributed equally to this work and are joint first authors. The views expressed are those of the author(s) and not necessarily those of the National Health Service (NHS), the National Institute for Health Research (NIHR), Economic and Social Research Council (ESRC) and Medical Research Council (MRC) or the Department of Health, Western Australia.

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: This work was funded by the National Basic Research Program of China (2014CB846104), the National Key Technology R&D Program (2015BAI13B01), and the National Natural Science Foundation of China (81301171). W.C. was funded by the Department of Health, Western Australia. S.V. was funded by the National Institute for Health Research (NIHR) Biomedical Research Centre at South London and Maudsley National Health Service (NHS) Foundation Trust and King’s College London. G.H.-J. is jointly funded by the Economic and Social Research Council (ESRC) and Medical Research Council (MRC) under the Lifelong Health and Wellbeing Cross-Council Programme initiative (ES/L002892/1).

Notes

Author Biographies

Lu Liu, PhD, is currently a research assistant professor in the Department of Child Psychiatry, Peking University Sixth Hospital/Institute of Mental Health. She received her PhD in psychiatry in 2012. Her major interests are in the neuropsychology and genetics of ADHD.

Wai Chen, MRCPsych, PhD, is a professor of child psychiatry in University of Western Australia, and the head of service, statewide Tier 4 Complex Attention and Hyperactivity Disorders Service (CAHDS), Department of Health, Western Australia. His research is in the areas of ADHD, neurodevelopmental disorders, emotional dysregulation, and their treatments. He also served as a peer reviewer for the DSM-5 Clinical and Public Health Committee (CPHC) during the Diagnostic and Statistical Manual of Mental Disorders (5th ed.; DSM-5) revision.

Silia Vitoratou, PhD, is currently a postdoctoral research fellow at the Department of Biostatics, Institute of Psychiatry, Psychology and Neuroscience, King’s College London. She received her PhD in statistics in 2013. Her major research interest is the use of latent variable models in methodological and applied psychometrics.

Li Sun, MD, PhD, is a professor in Peking University Sixth Hospital/Institute of Mental Health. She got her PhD in 2003. Her major interests are in neuropsychology and neurophysiology of ADHD.

Xiaoyan Yu, BS, is currently a master’s degree candidate in Peking University Sixth Hospital/Institute of Mental Health. Her research mainly focuses on neuropsychology and neuroimaging of ADHD.

Gareth Hagger-Johnson, PhD, is senior research associate, Institute of Epidemiology & Health, University College London. His research includes epidemiology, quantitative psychology, and statistics. He has worked on data from several large cohort studies. He is on the editorial board of three journals and is author/coauthor of three books on research methods, statistics, and data analysis in health and psychology.

Zhaomin Wu, BS, is currently a PhD candidate in Peking University Sixth Hospital/Institute of Mental Health. Her research mainly focuses on the neuroimaging and genetics of ADHD.

Li Yang, MD, PhD, is an associate professor in Peking University Sixth Hospital/Institute of Mental Health. She got her PhD in 2004. Her major interests are in pharmacogenomics and psychology of ADHD.

Qiujin Qian, MD, PhD, is a professor and also the director-in-chief of the Department of Child Psychiatry, Peking University Sixth Hospital/Institute of Mental Health. She got her PhD in 2002. Her major interests are in neuropsychology and neurophysiology of ADHD.

Yufeng Wang, MD, PhD, is a professor in Peking University Sixth Hospital/Institute of Mental Health. Her major interests are in therapeutics, neuropsychology, neurophysiology, and genetics of ADHD.

References

American Psychiatric Association. (1994). Diagnostic and statistical manual of mental disorders (4th ed.). Washington, DC: Author.

American Psychiatric Association. (2013). Diagnostic and statistical manual of mental disorders (5th ed.). Arlington, VA: American Psychiatric Publishing.

Barkley

R. A.

(1998). Attention-deficit hyperactivity disorder: A clinical workbook (2nd ed.). New York, NY: Guilford Press.

Barkley

R. A.

Fischer

(2010). The unique contribution of emotional impulsiveness to impairment in major life activities in hyperactive children as adults. Journal of the American Academy of Child & Adolescent Psychiatry, 49, 503-513.

Bentler

P. M.

(1990). Comparative fit indexes in structural models. Psychological Bulletin, 107, 238-246.

Bentler

P. M.

Bonett

D. G.

(1980). Significance tests and goodness of fit in the analysis of covariance structures. Psychological Bulletin, 88, 588-606.

Biederman

Spencer

T. J.

Petty

Hyder

L. L.

O’Connor

K. B.

Surman

C. B.

Faraone

S. V.

(2012). Longitudinal course of deficient emotional self-regulation CBCL profile in youth with ADHD: Prospective controlled study. Neuropsychiatric Disease and Treatment, 8, 267-276.

Brown

(2006). Confirmatory factor analysis for applied research. New York, NY: Guilford Press.

Browne

M. W.

(2001). An overview of analytic rotation in exploratory factor analysis. Multivariate Behavioral Research, 36, 111-150.

10.

Browne

M. W.

Cudeck

(1993). Alternative ways of assessing model fit. In Bollen

K. A.

Long

J. S.

(Eds.), Testing structural equation models (pp. 136-162). Newbury Park, CA: Sage.

11.

Burke

J. D.

Hipwell

A. E.

Loeber

(2010). Dimensions of oppositional defiant disorder as predictors of depression and conduct disorder in preadolescent girls. Journal of the American Academy of Child & Adolescent Psychiatry, 49, 484-492.

12.

Chronis-Tuscano

Molina

B. S.

Pelham

W. E.

Applegate

Dahlke

Overmyer

Lahey

B. B.

(2010). Very early predictors of adolescent depression and suicide attempts in children with attention-deficit/hyperactivity disorder. Archives of General Psychiatry, 67, 1044-1051.

13.

Conners

C. K.

(1973). Rating scales for use in drug studies with children. Psychopharmacology Bulletin, 9, 24-29.

14.

Epstein

M. H.

Cullinan

Gadow

K. D.

(1986). Teacher ratings of hyperactivity in learning-disabled, emotionally disturbed, and mentally retarded children. The Journal of Special Education, 20, 219-229.

15.

Furlong

M. J.

Fortman

J. B.

(1984). Factor analysis of the abbreviated Conners Teacher Rating Scale: Implications for the assessment of hyperactivity. Psychology in Schools, 21, 289-293.

16.

Gong

Cai

(1993). Wechsler Intelligence Scale for Children–Chinese Revision (C-WISC). Changsha, China: Hunan Maps Press.

17.

Goyette

C. H.

Conners

C. K.

Ulrich

R. F.

(1978). Normative data on revised Conners Parent and Teacher Rating Scales. Journal of Abnormal Child Psychology, 6, 221-236.

18.

Guan

Wang

Chen

Yang

Qian

. . . Wang

(2009). A high-density single-nucleotide polymorphism screen of 23 candidate genes in attention deficit hyperactivity disorder: Suggesting multiple susceptibility genes among Chinese Han population. Molecular Psychiatry, 14, 546-554.

19.

Hoelter

J. W.

(1983).The analysis of covariance structures: Goodness-of-fit indices. Sociological Methods & Research, 11, 325-344.

20.

Hulvershorn

L. A.

Mennes

Castellanos

F. X.

Di Martino

Milham

M. P.

Hummer

T. A.

Roy

A. K.

(2014). Abnormal amygdala functional connectivity associated with emotional lability in children with attention-deficit/hyperactivity disorder. Journal of the American Academy of Child & Adolescent Psychiatry, 53, 351-361.

21.

James

Lai

F. H.

Dahl

(2004). Attention deficit hyperactivity disorder and suicide: A review of possible associations. Acta Psychiatrica Scandinavica, 110, 408-415.

22.

Krieger

F. V.

Polanczyk

V. G.

Goodman

Rohde

L. A.

Graeff-Martins

A. S.

Salum

. . . Stringaris

(2013). Dimensions of oppositionality in a Brazilian community sample: Testing the DSM-5 proposal and etiological links. Journal of the American Academy of Child & Adolescent Psychiatry, 52, 389-400.e1.

23.

Mandy

Roughan

Skuse

(2014). Three dimensions of oppositionality in autism spectrum disorder. Journal of Abnormal Child Psychology, 42, 291-300.

24.

Margalit

(1983). Diagnostic application of the Conners Abbreviated Symptom Questionnaire. Journal of Clinical Child Psychology, 12, 355-357.

25.

Merwood

Chen

Rijsdijk

Skirrow

Larsson

Thapar

Asherson

(2014). Genetic associations between the symptoms of attention-deficit/hyperactivity disorder and emotional lability in child and adolescent twins. Journal of the American Academy of Child & Adolescent Psychiatry, 53, 209-220.

26.

Muthén

Christoffersson

(1981). Simultaneous factor analysis of dichotomous variables in several groups. Psychometrika, 46, 407-419.

27.

Muthén

du Toit

S. H. C.

Spisic

(1997). Robust inference using weighted least squares and quadratic estimating equations in latent variable modelling with categorical and continuous outcomes (Unpublished technical report). Retrieved from Mplus website: http://www.statmodel.com/wlscv.shtml

28.

Muthén

L. K.

Muthén

B. O.

(1998-2011). Mplus user’s guide (6th ed.). Los Angeles, CA: Author.

29.

Parker

J. D. A.

Sitarenios

Conners

C. K.

(1996). Abbreviated Conners’ Rating Scales revisited: A confirmatory factor analytic study. Journal of Attention Disorders, 1, 55-62.

30.

Posner

Rauh

Gruber

Gat

Wang

Peterson

B. S.

(2013). Dissociable attentional and affective circuits in medication-naive children with attention-deficit/hyperactivity disorder. Psychiatry Research, 213, 24-30.

31.

Reimherr

F. W.

Marchant

B. K.

Olsen

J. L.

Halls

Kondo

D. G.

Williams

E. D.

Robison

R. J.

(2010). Emotional dysregulation as a core feature of adult ADHD: Its relationship with clinical variables and treatment response in two methylphenidate trials. Journal of ADHD and Related Disorders, 1, 53-64.

32.

Reis

H. T.

Judd

C. M.

(2000). Handbook of research methods in social and personality psychology. New York, NY: Cambridge University Press.

33.

Seymour

K. E.

Chronis-Tuscano

Iwamoto

D. K.

Kurdziel

Macpherson

(2014). Emotion regulation mediates the association between ADHD and depressive symptoms in a community sample of youth. Journal of Abnormal Child Psychology, 42, 611-621.

34.

Skirrow

Asherson

(2013). Emotional lability, comorbidity and impairment in adults with attention-deficit hyperactivity disorder. Journal of Affective Disorders, 147, 80-86.

35.

Sobanski

Banaschewski

Asherson

Buitelaar

Chen

Franke

. . . Faraone

S. V.

(2010). Emotional lability in children and adolescents with attention deficit/hyperactivity disorder (ADHD): Clinical correlates and familial prevalence. Journal of Child Psychology and Psychiatry, 51, 915-923.

36.

Stringaris

Goodman

(2009). Longitudinal outcome of youth oppositionality: Irritable, headstrong, and hurtful behaviors have distinctive predictions. Journal of the American Academy of Child & Adolescent Psychiatry, 48, 404-412.

37.

L. Y.

Geng

Y. G.

Wang

Y. S.

Sun

Zhao

F. T.

Fan

(2006). Norm of ADHD Diagnostic Scale–Parent Version in Chinese urban children. Zhong Guo Shi Yong Er Ke Za Zhi, 21, 833-836.

38.

Surman

C. B.

Biederman

Spencer

Miller

C. A.

McDermott

K. M.

Faraone

S. V.

(2013). Understanding deficient emotional self-regulation in adults with attention deficit hyperactivity disorder: A controlled study. Attention Deficit Hyperactivity Disorder, 5, 273-281.

39.

Surman

C. B.

Biederman

Spencer

Yorks

Miller

C. A.

Petty

C. R.

Faraone

S. V.

(2011). Deficient emotional self-regulation and adult attention deficit hyperactivity disorder: A family risk analysis. American Journal of Psychiatry, 168, 617-623.

40.

Tillman

Geller

(2005). A brief screening tool for a prepubertal and early adolescent bipolar disorder phenotype. American Journal of Psychiatry, 162, 1214-1216.

41.

T. Y.

(1999). Rating scales for mental health. Beijing: Chinese Mental Health Journal.

42.

Yang

Wang

Y. F.

Qian

Q. J.

Biederman

Faraone

S. V.

(2004). DSM-IV subtypes of ADHD in a Chinese outpatient sample. Journal of the American Academy of Child & Adolescent Psychiatry, 43, 248-250.

43.

Yates

(1987). Multivariate exploratory data analysis: A perspective on exploratory factor analysis. Albany: State University of New York Press.

44.

Yüce

Zoroglu

S. S.

Ceylan

M. F.

Kandemir

Karabekiroglu

(2013). Psychiatric comorbidity distribution and diversities in children and adolescents with attention deficit/hyperactivity disorder: A study from Turkey. Neuropsychiatric Disease and Treatment, 9, 1791-1799.