What’s race got to do with it? Informant rating discrepancies in neuropsychological evaluations for children with ADHD

Abstract

Objective:

To evaluate race-based discrepancies in informant ratings and in rates of Attention-Deficit/Hyperactivity Disorder (ADHD) diagnosis among a clinically referred sample of Black and White children.

Method:

Demographic information and ratings of inattention, hyperactivity/impulsivity, and conduct were collected from caregivers and teachers as part of neuropsychological evaluations at an outpatient clinic. The final sample included 3,943 children (6–18 years), of which 70% were White and 30% were Black.

Results:

Teachers, but not caregivers, endorsed more inattentive symptoms and conduct problems for Black than for White children, irrespective of ADHD diagnostic status and socioeconomic status (SES), and after controlling for child sex, child age, and learning difficulties. Teachers endorsed more hyperactive/impulsive symptoms for Black children with ADHD of lower SES than for White children with these characteristics. Caregivers of Black children of higher SES reported fewer hyperactive/impulsive symptoms than caregivers of White children of higher SES. Despite differences in teachers’ ratings by race, diagnostic rates of ADHD in the context of neuropsychological evaluations were comparable for Black and White children.

Conclusions:

Consistent with previous literature, teachers endorsed more ADHD and conduct problems in Black children. Within our clinically referred sample, this may reflect teacher bias rather than actual prevalence differences by rafce, given that Black caregivers endorsed fewer or similar numbers of symptoms relative to White caregivers. This lack of racial disparities in rates of ADHD diagnosis is inconsistent with findings in community- and population-based samples, and reflects-possible benefit of the use of neuropsychological evaluations in diagnostic decision-making for ADHD.

Introduction

Racial disparities in the prevalence of ADHD

Attention-Deficit/Hyperactivity Disorder (ADHD) is one of the most common neurodevelopmental disabilities with rates of diagnosis on the rise (Davidovitch et al., 2017; Fairman et al., 2020), and recent prevalence estimates range from 7–15% (Danielson et al., 2018; Pastor et al., 2015; Rowland et al., 2015; Thomas et al., 2015; Visser et al., 2014). Research is mixed, however, on the variability in ADHD diagnostic rates by race. While ample research suggests that Black youth are under-diagnosed with ADHD compared to White peers (Bax et al., 2019; Coker et al., 2016; Morgan et al., 2013; 2014; Pastor & Reuben, 2005; Schneider & Eisenberg, 2006; Stevens et al., 2005; Visser et al., 2016), other studies suggest that Black youth are diagnosed with ADHD at rates similar to or higher than White youth (Angold et al., 2002; Collins & Cleary, 2016; Danielson et al., 2018; Pastor et al., 2015; Visser et al., 2014; Xu et al., 2018; Zablotsky & Alford, 2020). Moreover, some research suggests that the gap in ADHD diagnosis between Black and White children may be closing (Collins & Cleary, 2016; Zablotsky & Alford, 2020), yet other studies continue to find evidence pointing to the under-diagnosis of ADHD in Black children (e.g., Bax et al., 2019; Coker et al., 2016; Visser et al., 2016). The reasons for the variability in ADHD diagnostic rates by race remain unclear; however, mixed findings may be the result of methodological differences, as well as the differential impact of covariates, such as socioeconomic status (SES), on race. Despite inconsistencies in population- and community-based studies, rates of ADHD diagnosis by race have not been thoroughly explored in clinically referred samples.

It is also necessary to consider the intersection of SES and racial disparities in the diagnosis of ADHD. Although there is research that suggests that children from lower SES backgrounds are more likely to be diagnosed with ADHD than children from higher SES backgrounds (Bax et al., 2019; Pastor & Reuben, 2005; Rowland et al., 2018; Russell et al., 2015; Xu et al., 2018), the role of SES in the diagnosis and treatment of ADHD is complex. For example, children with medical assistance may be more likely to be diagnosedwithADHD, but may be less likely to be prescribed a stimulant medication compared to children with ADHD with commercial insurance (Stevens et al., 2004, 2005). Nonetheless, while research that has explicitly examined the intersection of race and SES in the diagnosis of ADHD is limited, studies have demonstrated that lower rates of ADHD diagnosis continue to persist for Black children even when accounting for SES (e.g., Morgan et al., 2013; 2014; Pastor & Reuben, 2005). Even though, in general, children with lower SES may be more likely to be diagnosed with ADHD, higher levels of parental education may increase rates of ADHD diagnosis for White children, but not for Black children (Zablotsky & Alford, 2020). Thus, disentangling race-based diagnostic discrepancies from those related to SES, as well as additional mediating factors, has proven challenging, yet such research is needed to shed light on inequities in service provision.

Comorbidity and ADHD

Upwards of 50–60% of children with ADHD also meet diagnostic criteria for disruptive behavior disorders, including oppositional Defiant Disorder (ODD) and Conduct Disorder (CD; Gillberg et al., 2004; Jensen et al., 2001; Visser et al., 2016). While research investigating racial disparities in ADHD comorbidities is limited, one study found that Black children aremore likely to have comorbid diagnoses of ADHD and ODD/CD, whereas White children aremore likely to have diagnoses of ADHD alone (Visser et al., 2016). Still, research on prevalence rates of ODD/CD in Black children is inconsistent (Angold et al., 2002; Bird et al., 2001; Cameron & Guterman, 2007; Costello et al., 2001; Feisthamel & Schwartz, 2009; Roberts et al., 2006).

Perhaps these mixed findings about the prevalence of ODD/CD in Black youth are influenced by factors embedded in structural racism (Cunningham et al., 2016; Fadus et al., 2020). First, Black children are disproportionately exposed to adverse childhood experiences that can contribute to neurodevelopmental and mental health disorders (e.g., Coker et al., 2009). Second, Black children are more likely to be misdiagnosed and/or receive more severe diagnoses (Awosan et al., 2011; Fadus et al., 2020; Feisthamel & Schwartz, 2009). Given that ADHD is likely under-diagnosed in Black youth, behavioral and academic difficulties may be interpreted as symptoms of ODD/CD, or simply attributed to willful misbehavior. The misattribution of symptoms may then lead parents, teachers, and society at large to punish Black children more often (Jacobsen et al., 2019), instead of providing appropriate intervention. Finally, although ADHD may be under-diagnosedin Black children, Black youthwith ADHD are incarcerated at higher rates than are White youth with ADHD (Soltis et al., 2017). Given the deleterious risks for poorer outcomes, the assessment and diagnosis of ADHD is arguably more complex and carries significantly more weight for Black children.

Informant ratings of ADHD symptoms

The Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5) diagnostic criteria for ADHD requires symptoms of inattention and/or hyperactivity/impulsivity to be present across two different settings (American Psychiatric Association (APA), 2013). Thus, clinicians often ascertain both caregiver and teacher ratings of ADHD symptoms, as well as related externalizing problems, when evaluating children for ADHD. Although both caregiver and teacher ratings are crucial parts of the ADHD diagnostic process, caregiver and teacher ratings of ADHD symptoms are not usually highly correlated with one another (Narad et al., 2015; Sollie et al., 2013; Willcutt et al., 2012; Wolraich et al., 2004). Measurement error is not a complete explanation for informant discrepancies (De Los Reyes, 2013; DuPaul et al., 2020). Rather, two alternative explanations have been proposed: (1) the unique perspective hypothesis, which maintains that differences in informant reports occur because caregivers and teachers have different perceptions of problem behavior (Antrop et al., 2002; Gomez, 2007; Mitsis et al., 2000; Wolraich et al., 2004), and (2) the situational specificity hypothesis, which suggests that discrepancies between informants reflect differences in expectations and demands in specific settings (De Los Reyes & Kazdin, 2005; Dirks et al., 2012; Gomez, 2007).

Caregiver ratings

Compared to teachers, caregivers tend to endorse more symptoms of ADHD (Antrop et al., 2002; Malhi et al., 2008; Murray et al., 2018; Takeda et al., 2020). Additionally, research suggests that discrepancies between caregiver and teacher ratings are larger for children from diverse racial/ethnic groups (Lau et al., 2004; Takeda et al., 2020). Although caregivers, in general, tend to endorse more ADHD symptoms compared to teachers, some research indicates that caregivers of Black children endorse fewer ADHD symptoms than do their teachers (Harvey et al., 2013; Kang & Harvey, 2020; Lawson et al., 2017). Accordingly, what factors influence caregivers’ and teachers’ ratings and whether this leads to over- or under-estimation of ADHD symptoms warrants further exploration.

Teacher ratings

Informant ratings, especially those from teachers, are often of great value to clinicians because they provide insight about the child’s behavior in a setting other than the home. While informant ratings are often considered integral in the diagnosis of ADHD, studies using community- and population-based samples have found that teachers endorse higher rates of ADHD symptoms and related externalizing behaviors in Black children compared to White children (DuPaul et al., 2014, 2016; Epstein et al., 1998). In one sample of clinically referred children, researchers also found similar results (Lawson et al., 2017). None of these studies, however, examined differences in rates of ADHD diagnosis by race.

Trends in higher reports of problem behavior for Black children hold true across contexts; teachers rate Black children as experiencing higher levels of executive dysfunction (e.g., lower frustration tolerance, greater difficulty staying on task) in preschool settings (Sbarra & Pianta, 2001) and throughout early elementary school (Garcia et al., 2019), as well as engaging in significantly more disruptive behaviors requiring disciplinary action when compared to White peers (Jacobsen et al., 2019; Kaufman et al., 2010; Riddle & Sinclair, 2019). Interestingly, Black teachers have been found to rate fewer behavioral and academic problems for Black children (Driessen, 2015; Redding, 2019). However, racial/ethnic matching between teachers and students not been found to neither reduce racially discrepant punitive disciplinary practices nor improve academic outcomes for Black children (Bradshaw et al., 2010; Driessen, 2015; Redding, 2019). Researchers have also found that White teachers with more negative racial attitudes were more likely to endorse higher rates of ADHD symptoms in Black children compared to caregivers, even when they were both shown identical videos of child behavior (Kang & Harvey, 2020). Consequently, these higher rates were not attributable to true differences in ADHD presentation in the classroom environment (Kang & Harvey, 2020). While we may expect behavioral ratings to function differently across race and/or culture, research has found that ADHD symptom report variability is not due to item-level measurement variance across child characteristics (DuPaul et al., 2020). The lack of evidence for item-level measurement variance across child characteristics is consistent with research suggesting that rater variability may be due to distorted perceptions shaped by systemic biases, stereotypes, and static negative attributions of Black children (De Los Reyes & Kazdin, 2005; Kang & Harvey, 2020; Kunesh & Noltemeyer, 2015). Thus, given the evidence that teacher ratings may not accurately represent Black children’s behavior, further investigation can aid neuropsychologists in conceptualizing their diagnostic procedures for Black children suspected of having ADHD.

Current study

The purpose of this study was twofold. First, we compared caregiver and teacher ratings of symptoms of inattention, hyperactivity/impulsivity, and conduct problems between Black and White children who did and did not receive diagnoses of ADHDduring neuropsychological evaluation, as well as examined the interactions among race, SES, and diagnostic status in informant ratings. Based on previous research suggesting that teachers endorse more externalizing symptoms for Black children, we hypothesized that Black children, both those with and without diagnoses of ADHD, would have significantly greater symptoms of inattention, hyperactivity/impulsivity, and conduct problems endorsed by teachers on rating scales as compared to their White peers. Given limited evidence to support differences in caregiver ratings by race, we hypothesized that there would not be significant differences between caregivers’ ratings of ADHD symptoms and conduct problems for White and Black children. Second, limited research has investigated racial disparities in rates of ADHD diagnosis in clinically referred samples; thus, we aimed to examine rates of ADHD diagnosis between Black and White children in our clinically referred sample, while also considering the role of other sociodemographic variables, such as child sex, child age, learning difficulties, and SES. Given previous literature pointing to the under-diagnosis of Black children with ADHD, our goal was to determine whether racial disparities in ADHD diagnosis held true at our large, urban neuropsychological outpatient clinic. Ultimately, examining this diagnostic process from a social justice lens can help neuropsychologists better understand the role of race in ratings of inattention, hyperactivity/impulsivity, and conduct problems and aid in equitable treatment planning and recommendations.

Method

Sample

Data from caregiver and teacher ratings that were collected between December 2011 and October 2020 were obtained as part of routine clinical care in a neuropsychology outpatient clinic at a large, urban academic medical center. Clinicians routinely enter assessment data into a clinical database via the secure electronic health record, and these data are maintained securely by the hospital’s information Systems department. upon approval by the institutional Review Board (IRB), under a waiver of consent, a de-identified dataset was constructed from the clinical database of patients between age 6–18 years (mean age = 10.55 years, SD = 3.01), who were identified by their caregiver as either Black or White and for whom caregiver and teacher symptom ratings, as well as clinical diagnoses, were available.

The final sample included 3,943 children, of which 70% were White (n = 2,754) and 30% were Black (n = 1,189). Approximately 44% (n = 1,724) had a primary or secondary billing diagnosis of ADHD, predominantly inattentive presentation; ADHD, predominantly hyperactive/impulsive presentation; or ADHD, combined presentation. of those who were not diagnosed with ADHD, 37.2% had primary billing diagnoses of other mental health disorders and neurodevelopmental disabilities (e.g., anxiety, depression, language disorders) and 19.1% had primary billing diagnoses of medical conditions (e.g., epilepsy, oncologic diseases, encephalopathy). Please see Table 1 for a comparison of sample demographics by race.

Table 1.

Sample demographic characteristics.

		White		Black
	N	(n = 2,754)		(n = 1,189)		χ2
Child sex (% male)	3943	62.5%		65.7%		3.57
Diagnostic status (% with ADHD)	3943	43.0%		45.4%		1.79
Insurance type (% with commercial insurance)	3336	83.1%		57.8%		231.88***
		M	SD	M	SD	t
Child age (years)	3943	10.49	3.02	10.70	2.97	2.10
CLDQ Reading1	3943	2.87	1.28	2.96	1.25	2.06*
CLDQ Math1	3943	2.92	1.22	3.11	1.22	4.48***

Note.

^***p<.001,

^*p<.05.

¹CLDQreading and math scores are represented by the mean of informants’ responses to Likert scale items.

ADHD = Attention-Deficit/Hyperactivity Disorder; CLDQ = Colorado Learning Difficulties Questionnaire. Comparisons between groups for categorical variables were conducted using chi-square tests (χ²) and for continuous variables were conducted as independent-samples t-tests.

Measures

The measures used in the current study were from a series of pre-visit symptom report measures completed by caregivers and teachers as part of routine care (Zabel et al., 2020).

ADHD-Rating scale-fifth edition, home and school versions (ADHD-RS-5; DuPaul et al., 2016)

The ADHD-RS-5 Home Version is a caregiver-completed rating of symptoms of ADHD that closely corresponds to the DSM-5 diagnostic criteria. The ADHD-RS-5 School Version is identical to the Home Version but is completed by the teacher. The measure includes subscales of inattention (nine items) and Hyperactivity/Impulsivity (nine items). The questions were rated on a four-point Likert scale based on how frequently symptoms occur (0 = “never,” 1 = “sometimes,” 2 = “often,” 3 = “very often”). Consistent with previous literature (e.g., Curchack-Lichtin et al., 2014), a symptom was considered to be endorsed if the informant indicated that the behavior occurred “often” or “very often,” and symptom counts were used in all statistical analyses. In prior work, this measure has demonstrated strong internal consistency, as well as adequate criterion-related, discriminant, and predictive validity (DuPaul et al., 2016). In the current sample, internal consistency via Cronbach’s alpha was excellent (teacher-report inattention α = .93, Hyperactivity/Impulsivity α = .93; parent-report inattention α = .90, Hyperactivity/Impulsivity α = .91). Internal consistency metrics were identical when raters were reporting on Black versus White children.

NICHQ Vanderbilt assessment scales, parent and teacher informants (Wolraich et al., 2003)

The Vanderbilt Assessment Scales assess symptoms related to prevalent pediatric conditions, including ADHD, ODD, CD, anxiety, and depression. For the purposes of this study, a subset of eight items that were the same across the caregiver and teacher forms were used to measure symptoms of ODD and CD (parent α = .94; teacher α = .93; similarly excellent Internal consistency for both Black and White children). These items were rated on a four-point Likert scale based on how frequently symptoms occur (0 = “never,” 1 = “sometimes,” 2 = “often,” 3 = “very often”). Symptoms were considered present if informants endorsed that behaviorsoccurred “often” or “very often,” and these symptom counts were used in statistical analyses. The Vanderbilt Assessment Scales have strong Internal consistency, test-retest reliability, and predictive validity (Wolraich et al., 2013).

Colorado learning difficulties questionnaire (CLDQ; Willcutt et al., 2011)

The CLDQ is a caregiver-reported measure of learning difficulties. For this study, only the Reading (six items) and Math (five items) subscales were utilized. Caregivers rated each question on a five-point Likert scale, indicating how well the statement described their child (1 = “never/not at all,” 2 = “rarely/a little,” 3 = “sometimes,” 4 = “frequently/quite a bit,” 5 = “always/a great deal”). This measure has demonstrated strong Internal consistency and predictive validity within both community and clinical samples (Patrick et al., 2013; Willcutt et al., 2011); strong Internal consistency was similarly observed with these data overall (CLDQ Reading α = .93, CLDQ Math α = .92), as well as when examined with Black and White children separately.

Covariates

Child sex

Child sex (i.e., male or female), as reported by the caregiver, was obtained from the electronic medical record system.

Child age

Child age was calculated using the child’s birthdate and the date that the caregiver completed the pre-visit questionnaire.

Socioeconomic status (SES)

Insurance type (n = 3,336) was obtained from the medical record and used as a proxy for SES. Binary categories included medical assistance (n = 793; 23.8%) and commercial insurance (n = 2543; 76.2%).

Learning difficulties

Learning difficulties were assessed via caregiver-report on the CLDQ (n = 3,943). Two variables consisting of the mean scores for reading difficulties and math difficulties, respectively, on the CLDQ were calculated and used as covariates.

Statistical analyses

To preliminarily investigate whether caregivers and teachers differed in their ratings of Black and White children who did and did not receive a diagnosis of ADHD, initial independent and paired-sample t-tests were conducted (see Tables 2 and 3). Cohen’s d and Hedges’ g were calculated as effect size measures based on equal and unequal group size, respectively.

Table 2.

Descriptive statistics and comparisons by child race.

	N	Black Children M (SD)	White Children M (SD)	T-test	Hedges’ g
Children with ADHD
Caregiver IA	1724	6.33 (2.56)	6.10 (2.58)	1.68	.09
Caregiver HI	1724	3.96 (3.01)	4.04 (2.94)	.52	.03
Caregiver Conduct	1711	2.72 (2.53)	2.83 (2.36)	.38	.05
Teacher IA	1724	5.07 (3.15)	4.54 (3.07)	3.38***	.18
Teacher HI	1724	3.17 (3.14)	2.60 (2.92)	3.53***	.19
Teacher Conduct	1724	1.21 (2.22)	.75 (1.70)	4.30***	.25
Children without ADHD
Caregiver IA	2219	4.72 (2.98)	4.56 (3.00)	1.21	.06
Caregiver HI	2219	2.55 (2.65)	2.47 (2.55)	.67	.03
Caregiver Conduct	2206	2.25 (2.38)	2.16 (2.24)	.83	.04
Teacher IA	2219	4.08 (3.13)	3.19 (3.04)	6.26***	.29
Teacher HI	2219	2.14 (2.78)	1.57 (2.36)	4.55***	.23
Teacher Conduct	2219	.94 (1.99)	.46 (1.31)	5.66***	.31

Note.

^***p<.001.

ADHD = Attention-Deficit/Hyperactivity Disorder; IA = inattentive symptoms; HI = hyperactive/impulsive symptoms. Within-informant race-based comparisons were conducted as independent-samples t-tests. Hedges’ g was calculated due to unequal group size.

Table 3.

Diagnostic group comparisons by child race for ADHD and conduct symptoms.

	N	Children with ADHD M (SD)
Without ADHD M (SD)	T-test	Hedges’ g	Children
Black children
Caregiver IA	1189	6.33 (2.56)	4.72 (2.98)	9.97***	.57
Caregiver HI	1189	3.96 (3.01)	2.55 (2.65)	8.50***	.50
Caregiver Conduct	1183	2.72 (2.53)	2.25 (2.38)	3.25**	.19
Teacher IA	1189	5.07 (3.14)	4.09 (3.13)	5.40***	.32
Teacher HI	1189	3.17 (3.14)	2.14 (2.78)	5.92***	.35
Teacher Conduct	1189	1.21 (2.22)	.94 (1.99)	2.21*	.13
White children
Caregiver IA	2754	6.10 (2.58)	4.56 (3.00)	14.51***	.55
Caregiver HI	2754	4.04 (2.93)	2.47 (2.55)	14.73***	.58
Caregiver Conduct	2734	2.83 (2.36)	2.16 (2.24)	7.54***	.29
Teacher IA	2754	4.53 (3.07)	3.19 (3.04)	11.41***	.44
Teacher HI	2754	2.60 (2.92)	1.57 (2.36)	9.95***	.39
Teacher Conduct	2754	.75 (1.70)	.46 (1.31)	4.89***	.20

Note.

^***p<.001,

^**p<.01,

^*p<.05.

ADHD = attention-Deficit/hyperactivity Disorder; IA = inattentive symptoms; HI = hyperactive/impulsive symptoms. Within-informant diagnostic-group comparisons were conducted as independent-samples t-tests. Hedges’ g was calculated due to unequal group size.

Next, three-way between-subjects univariate analyses of variance (ANOVAs) were conducted in order to examine potential interactions among race, diagnostic status, and SES (i.e., insurance type) on caregiver and teacher report of inattentive, hyperactive/impulsive, and conduct symptoms, as well as the main effect of race. Three-way between-subjects models included a 2(race: Black, White) by 2(diagnosis: with ADHD, without ADHD) by 2(insurance: medical assistance, commercial) model with child age, child sex, and learning difficulties¹ included as covariates in all ANOVA analyses. Outcomes included caregiver- and teacher-reported inattentive, hyperactive/impulsive, and conduct symptoms (i.e., three models were run with caregiver-reported symptoms and three with teacher-reported symptoms as outcomes). Significant interactions by race were probed further using t-tests and separate ANOVAs split by characteristics (e.g., with and without an ADHD diagnosis), as appropriate.

Lastly, to investigate differences in the rates of ADHD diagnosis between Black and White children, an initial chi-square test was conducted. A follow-up binary logistic regression analysis was run with insurance type, child age, child sex, and learning difficultiesin order toevaluate the relationship between race and diagnostic status in light of these important covariates.

Results

Caregiver ratings of ADHD symptoms and conduct problems

Inattentive symptoms

For inattentive symptoms as reported by caregivers, there were no significant interactions by child race, or main effects of child race. That is, regardless of ADHD diagnostic status, and with child sex, child age, and learning difficulties accounted for as covariates, caregivers of White and Black children did not differ in their ratings of inattentive symptoms; see Table 4 for caregiver-based analyses.

Table 4.

Caregiver report ANOVAs with child race, diagnostic status, and insurance type.

	df	Inattentive Symptoms (N = 3336)			Hyperactive/Impulsive Symptoms (N = 3336)			Conduct Symptoms (N = 3317)
		MS	F	p	MS	F	p	MS	F	p
(Intercept)	1	1684.06	230.35	<.001	3990.01	581.05	<.001	1712.61	325.38	<.001
Child age	1	2.64	0.36	.55	1317.49	191.86	<.001	152.77	29.03	<.001
Child sex	1	247.18	33.81	<.001	182.24	26.54	<.001	24.50	4.66	.03
Reading difficulties	1	11.40	1.56	.21	0.27	0.04	.84	38.05	7.23	.01
Math difficulties	1	1679.39	229.71	<.001	53.35	7.77	.01	7.77	1.48	.22
Race	1	0.30	0.04	.84	17.89	2.61	.11	20.05	3.81	.05
Insurance Type	1	125.50	17.17	<.001	465.66	67.81	<.001	417.06	79.24	<.001
ADHD dx	1	1056.58	144.52	<.001	870.87	126.82	<.001	202.56	38.48	<.001
Race by insurance type	1	17.43	2.38	.12	43.90	6.39	.01	26.95	5.12	.02
Race by ADHD dx	1	3.20	0.44	.51	1.00	0.15	.70	2.58	0.49	.48
Insurance type by ADHD dx	1	0.00	0.00	.99	15.23	2.22	.14	11.05	2.10	.15
Race by insurance type by ADHD dx	1	0.35	0.05	.83	4.45	0.65	.42	0.35	0.07	.80
Error	3324	7.31			6.87			5.26

Note. ADHD = Attention-Deficit/Hyperactivity Disorder; dx = diagnosis. For insurance: 0 = commercial insurance; 1 = medical assistance. For race: 0 = White; 1 = Black. ADHD dx: 0 = no diagnosis of ADHD; 1 = diagnosis of ADHD.

Hyperactive/impulsive symptoms

With regard to caregiver-reported hyperactive/impulsive symptoms, there was a main effect of race, which was qualified by an interaction between race and insurance type, F(1,3324) = 6.39, p = .01. Caregivers of Black children with commercial insurance reported fewer hyperactive/impulsive symptoms than caregivers of White children with commercial insurance, t(2541) = 3.28, p = .001, regardless of ADHD diagnostic status; whereas, caregivers of children with medical assistance reported similar levels of hyperactive/impulsive symptoms, regardless of race, t(791) = −.19, p = .85.

Conduct problems

An interaction was found between race and insurance type, F(1,3305) = 5.12, p = .02. Post-hoc analysis of this interaction suggested that caregivers of Black children with commercial insurance reported fewer conduct problems than caregivers of White children with commercial insurance, t(2530) = 3.63, p<.001, regardless of diagnostic status. As with hyperactive/impulsive symptoms, families with medical assistance reported similar levels of conduct problems, regardless of race and diagnostic status, t(783) = .06, p = .95.

Teacher ratings of ADHD symptoms and conduct problems

Inattentive symptoms

For teacher endorsement of inattentive symptoms, there were no significant interactions, but there was a significant main effect of race, such that teachers rated Black children as having more inattentive symptoms than White children. That is, regardless of child insurance type, child age, child sex, learning difficulties, and a child’s diagnosis of ADHD, teachers rated Black children as having more inattentive symptoms than White children; see Table 5 for teacher-based analyses.

Table 5.

Teacher report ANOVAs with child race, diagnostic status, and insurance type.

	df	Inattentive Symptoms (N = 3336)			Hyperactive/Impulsive Symptoms (N = 3336)			Conduct Symptoms (N = 3336)
		MS	F	p	MS	F	p	MS	F	p
(Intercept)	1	3536.40	404.85	<.001	3339.53	505.85	<.001	549.41	227.68	<.001
Child age	1	827.47	94.73	<.001	1041.01	157.69	<.001	143.22	59.35	<.001
Child sex	1	896.17	102.59	<.001	558.80	84.64	<.001	71.50	29.63	<.001
Reading difficulties	1	0.37	0.04	.84	67.96	10.29	.001	16.63	6.89	.01
Math difficulties	1	666.87	76.34	<.001	0.11	0.02	.90	6.54	2.71	.10
Race	1	69.88	8.00	.01	67.77	10.26	.001	63.58	26.35	<.001
Insurance Type	1	13.46	1.54	.22	74.21	11.24	.001	104.56	43.33	<.001
ADHD dx	1	383.84	43.94	<.001	317.16	48.04	<.001	15.91	6.59	.01
Race by insurance type	1	44.45	5.09	.02	11.01	1.67	.20	0.93	0.38	.54
Race by ADHD dx	1	14.66	1.68	.20	7.62	1.15	.28	0.04	0.02	.90
Insurance type by ADHD dx	1	16.75	1.92	.17	0.00	0.00	.99	0.00	0.00	.99
Race by insurance type by ADHD dx	1	0.97	0.11	.74	27.67	4.19	.04	0.40	0.17	.68
Error	3324	8.74			6.60			2.41

Note. ADHD = attention-Deficit/hyperactivity Disorder; dx = diagnosis. For insurance: 0 = commercial insurance; 1 = medical assistance. For race: 0 = White; 1 = Black. ADHD dx: 0 = no diagnosis of ADHD; 1 = diagnosis of ADHD.

Hyperactive/impulsive symptoms

A significant three-way interaction emerged between child race, child diagnostic status, and insurance type with regard to teacher endorsement of hyperactive/impulsive symptoms, F(1,3324) = 4.19, p = .04. To analyze this interaction, a 2(race) by 2(insurance type) ANOVA split by diagnostic status indicated a significant interaction between race and insurance type only within the group of children diagnosed with ADHD, F(1,1511) = 5.97, p = .015. Probed further via t-tests, for children with a diagnosis of ADHD (but not for children without a diagnosis of ADHD), Black children with medical assistance were rated as having more hyperactive/impulsive symptoms than White children with medical assistance, t(412) = −2.84, p = .005.

Conduct problems

When teacher-reported conduct problems were examined with race, diagnostic status, and insurance type, there were no significant interactions, but there was a main effect of race, F(1,3324) = 26.35, p<.001, such that Black children were rated as having more conduct problems, regardless of their diagnostic status, insurance type, age, sex, or learning difficulties.

Caregiver-teacher comparisons

In line with previous literature, caregivers endorsed more symptoms of inattention, as well as hyperactivity/impulsivity, than did teachers for the same children, paired-samples t(3942) = 21.97 and t(3942) = 19.75, respectively, p<.001 for both (Figure 1). This pattern was consistent regardless of ADHD diagnostic status and child race. Caregivers also reported more conduct problems than teachers for the same children, t(3916) = 43.38, p<.001, and this was similarly consistent, regardless of child race or diagnostic status (Figure 2). Both caregivers and teachers endorsed more inattentive, hyperactive/impulsive, and conduct symptoms for children diagnosed with ADHD than for children who were not diagnosed with ADHD, p<.001 for all comparisons, Hedge’s g range = 0.31–0.44; see Tables 2 and 3.

Figure 1.

Teacher and caregiver ratings of inattentive and hyperactive/impulsive symptoms by race, insuance type, and diagnostic status. Note. ADHD = Attention-Deficit/Hyperactivity Disorder; IA = inattentive symptoms; HI = hyperactive/impulsive symptoms; MA = medical assistance; Commercial = commercial insurance. n = 201 Black children with ADHD with MA; n = 230 White children with ADHD with MA; n = 180 Black children without ADHD with MA; n = 182 White children without ADHD with MA; n = 247 Black children with ADHD with commercial insurance; n = 837 White children with ADHD with commercial insurance; n = 275 Black children without ADHD with commercial insurance; n = 1184 White children without ADHD with commercial insurance. Symptoms were endorsed by caregivers and teachers on the ADHD-Rating Scale-Fifth Edition (ADHD-RS-5) and the Vanderbilt Assessment Scales.

Figure 2.

Teacher and caregiver ratings of conduct symptoms by race and insurance type. Note. MA = medical assistance; Commercial = commercial insurance. n = 381 Black children with MA; n = 412 White children with MA; n = 522 Black children with commercial insurance; n = 2021 White children with commercial insurance. Symptoms were endorsed by caregivers and teachers on the ADHD-Rating Scale-Fifth Edition (ADHD-RS-5) and the Vanderbilt Assessment Scales.

Rates of ADHD diagnosis

Rates of ADHD diagnoses within our clinic sample did not differ by child race (B = −0.09, SE B = 0.07, χ² = 1.79, p = .18), with 45% (n = 539 of 1189) of Black children receiving a diagnosis of ADHD compared to 43% (n = 1185 of 2754) of White children. These results suggest that, overall, Black and White children within our clinic received ADHD diagnoses at similar rates.This finding was maintained when covariates were added to the model, indicating that once insurance type and other important covariates were accounted for, Black and White children continued to be similarly likely to receive a diagnosis of ADHD.

Discussion

Caregiver ratings of ADHD symptoms

In this clinically referred sample, our findings suggest that caregivers of Black children with commercial insurance endorsed fewer hyperactive/impulsive symptoms and conduct problems than caregivers of White children, while caregivers of children with medical assistance endorsed similar levels of hyperactive/impulsive symptoms and conduct problems, regardless of race and ADHD diagnostic status. Additionally, caregiver ratings of inattentive symptoms did not differ based on race, diagnostic status, or insurance type. These findings were consistent after accounting for childage, childsex, and academic difficulties. Altogether, these findings suggest that caregivers of Black children with higher SES may be less likely to endorse externalizing symptoms for their children than caregivers of White children.

While some research has shown that caregivers of Black children are less likely to endorse behavioral problems (Harvey et al., 2013; Kang & Harvey, 2020; Lawson et al., 2017), extant literature exploring the intersection of race and SES on ADHD symptom ratings is scarce. Given our findings, we hypothesize that caregivers of Black children with higher SES are more likely to under-report hyperactive/impulsive symptoms and conduct problems due to an increased awareness of historic systemic racial biases for Black children (e.g., school disciplinary actions or involvement with the juvenile justice system; Blumstein, 2015; Jacobsen et al., 2019; Riddle & Sinclair, 2019). In one study, caregivers of Black boys reported fewer ADHD symptoms relative to teachers (Kang & Harvey, 2020), which may have been due to caregivers of Black children recognizing the unique challenges that Black men and boys face today, such as being perceived as threats (Todd et al., 2016). nonetheless, in this same study, where caregivers and teachers watched identical videos of children and then rated their ADHD symptoms, caregivers of Black children rated White children and Black girls similarly to White caregivers and teachers, suggesting that Black caregivers do not under-report ADHD symptoms in general (Kang & Harvey, 2020). Additionally, based on research that suggests that Black families are less likely to have access to care in their communities (Bax et al., 2019; Miller et al., 2009), caregivers of Black children with higher SES may be more likely to under-report behavioral concerns, given an increased availability of services relative to that of caregivers of Black children with lower SES, who may feel the need to be more transparent, or to over-emphasize, symptom severity in order to access necessary services. Alternatively, it is possible that Black children with higher SES had ADHD symptoms that were less impactful, such that lower symptom endorsement may be explained by less impairing symptoms based on the context in which they occurred.

Teacher ratings of ADHD symptoms

Our findings indicate that teachers endorsed more ADHD symptoms and conduct problems for Black children compared to White children. Although SES, diagnostic status, and other covariates did not influence teachers’ ratings of inattentive symptoms and conduct problems above and beyond race, teachers endorsed morehyperactive/impulsive symptoms for Black children only with a diagnosis of ADHD and medical assistance. These notable irregularities in informant data are consistent with previous literature suggesting that teachers endorse significantly more ADHD symptoms and behavioral issues in Black children compared to White children (e.g., DuPaul et al., 2014, 2016; Kang & Harvey, 2020), despite evidence suggesting that ADHD symptom items function similarly for Black and White children, irrespective of child and family characteristics (DuPaul et al., 2020). Our differences in teacher ratings between Black and White children resulted in small effects (Hedges’ g = .18–.31), which is generally consistent with previous literature that also found small to medium effects in community-based samples (e.g., DuPaul et al., 2014, 2016; Kang & Harvey, 2020). Given that race alone did not explain teachers’ over-endorsement of hyperactive/impulsive symptoms for Black children, our findings also highlight the potential roles that SES and diagnostic status, as well as race, play in teachers’ perceptions of child behavior. Even further, while differences in informant reports are expected due to setting variability (Dumenci et al., 2011), these discrepancies suggest that teacher ratings may be susceptible to teachers’ appraisal of child characteristics (e.g., race, SES) and may be shaped by systemic biases and negative racial attitudes towards Black children (De Los Reyes & Kazdin, 2005; Kang & Harvey, 2020; Kunesh & Noltemeyer, 2015).

Notably, teacher endorsement of conduct problems remained significant for Black children with and without ADHD, regardless of SES. Discrepancies in ratings for conduct problems in Black children are consistent with a growing body of literature establishing that Black youth are not only disciplined at significantly higher rates than White children (Jacobsen et al., 2019; Kaufman et al., 2010; Kinsler, 2011; Nicholson et al., 2009; Riddle & Sinclair, 2019), but are also more likely to be viewed as engaging in problematic behaviors and face harsher punishments when compared to their White counterparts (Jacobsen et al., 2019; Kinsler, 2011; Riddle & Sinclair, 2019).

Rates of ADHD diagnosis

Results suggest that rates of ADHD diagnosis between Black and White children in our sample were similar, even when accounting for the role of covariates. One interpretation of these findings is that clinicians in our department under-diagnosed ADHD in Black children, and teachers’ ratings of Black children’s symptoms were more accurate compared to their caregivers’ ratings. To this point, we found that caregivers endorsed significantly more symptoms than teachers, regardless of race, which is consistent with research suggesting that caregivers endorse more symptoms of ADHD (Antrop et al., 2002; Malhi et al., 2008; Murray et al., 2018; Takeda et al., 2020). Despite possible discrepancies between teacher and caregiver ratings, however, other lines of research have also found that caregiver ratings of ADHD symptoms yield similar diagnostic accuracy as do teacher ratings (Bied et al., 2017).

Accordingly, another explanation of our findings is that rates of ADHD diagnosisin our clinically referred sample were, in fact, similar between Black and White children, reflecting accurate diagnoses by clinicians.These findings are inconsistent with population- and community-based studies that posit the under-diagnosis of ADHD for Black children, andare more in line with recent research suggesting that the gap in diagnostic disparities is closing (Collins & Cleary, 2016; Zablotsky & Alford, 2020). Evaluations at our clinic, which typically involve thorough clinical interviews, may allow clinicians to connect withpatients and families in a way that is different fromthat of teachers in a school environment. In support of this hypothesis, semi-structured clinical interviews tend to be less sensitive to bias and are accepted as an important assessment method for the evaluation of ADHD symptoms and their impairment across settings (Pelham et al., 2005; Pliszka, 2007). The act of clinical interviewing may also promote perspective taking, which can reduce implicit bias and improve interactions between people of differing races (Todd et al., 2011). Given that neuropsychologists at our clinic regularly work closely with Black families to overcome barriers, advocate for their children, and access services, neuropsychological evaluations at our institution may help buffer potentially malinformed diagnoses on the basis of race. Further, research suggests that clinicians’ diagnostic decisions, regardless of evidence, tend to align with the referral source, or person who initiates clinical services (i.e., often caregivers), relative to other informants’ reports(De Los Reyes et al., 2015). This combination of thorough clinical interviews and perspective taking, along with clinicians’ propensity to align with referring caregivers, may help explain why rates of ADHD diagnosis between Black and White children in our sample were similar.

Importantly, the current study did not specifically address the role of neuropsychological evaluations in ADHD diagnosis, but it is also possible that the multimodal nature of neuropsychological evaluations contributed to the similar rates of ADHD diagnosis between Black and White children in this sample. In contrast to evaluations for ADHD conducted in other pediatric settings (e.g., primary care), neuropsychological evaluations for ADHD involve the use of not only symptom ratings from multiple informants, but also clinical interview, behavioral observations, and, as necessary, performance-based measures to assess skills across domains (Fine et al., 2018; Mahone & Slomine, 2008). While the utility of performance-based measures in the diagnosis of ADHD is contested (Barkley, 2019; Mapou, 2019), neuropsychological evaluations permit for the triangulation of data from multiple sources and across multiple modalities, which may not only promote better outcomes for all children (Pritchard et al., 2014), but also help mitigate biases that can impact behaviorally-based diagnoses, like ADHD.

Limitations and future directions

While our study adds to the existing literature and discourse on racial health disparities and biases in diagnostic tools for neurodevelopmental disorders, it is not without limitations. As our study is based on a clinically referred sample, we do not have comparative data from children who were not referred for neuropsychological evaluation. Given this constraint, the expansion of outcome variables and post-visit data will be invaluable for future research to further elucidate the relationship between the lower prevalence of ADHD in Black children in the context of teachers’ higher endorsement of ADHD symptoms for Black children. Of note, children were only included in our sample if both their caregiver and teacher completed ratings, making selection into our sample non-random. There are many possible reasons why children seen in our department, and those excluded from our sample, would not have had teacher ratings (e.g., caregiver chose not to ask teacher to complete ratings; teacher did not complete ratings due to limited time), although these reasonscan be neither fully quantified nor measured. In addition, we did not have access to variables on teacher or school characteristics, but we recognize that the school environment (e.g., culture, demographics, administration, funding), as well as teacher characteristics (e.g., race/ethnicity, years of experience, SES, age, gender, professional training), likely influence a teacher’s appraisal of child behavior and would be notable covariates for future investigation. For example, racial/ethnic matching between teachers and students has been found to have ameliorative impacts for children of color in suburban schools, but have neutral to negative impacts for children of color in urban or rural environments, respectively (Jung, 2020). Additionally, while research investigating the role of racial/ethnic matching between teachers and students suggests that Black teachers rate fewer behavioral and academic problems for Black children, this has not been demonstrated within or across other racial/ethnic groups (e.g., Latinx, Asian American, White; Driessen, 2015; Redding, 2019), indicating that teacher race may be particularly salient in behavioral ratings for Black children. However, despite the potential ameliorative effects of racial/ethnic matching on ratings for Black children, racial/ethnic matching neither reduces racially discrepant punitive disciplinary practices for Black children (Bradshaw et al., 2010) nor improves academic performance for Black children (Driessen, 2015; Redding, 2019). As such, teachers’ appraisal of Black children’s behavior is significant; teachers’ behavior ratings of Black children have been shown to be predictive of grade retention recommendations (Mattison et al., 2018), and office disciplinary referrals for Black children are linked to more exclusionary discipline practices (Jacobsen et al., 2019). While analyzing teacher and school characteristics may help further contextualize teachers’ over-endorsement of ADHD symptoms in Black children, the utility and impact of informant ratings on outcomes for Black children also requires further inquiry.

Moreover, our study is limited in its definition of SES. While our proxy for SES (i.e., insurance type) helped explain the variance in teachers’ over-endorsement of hyperactive/impulsive symptoms of Black children with ADHD, it is unclear why this interaction was not found for inattentive symptoms or conduct problems. Similarly, our results suggest that caregivers of Black children with commercial insurance under-endorsed hyperactive/impulsive symptoms and conduct problems in their children regardless of diagnostic status; however, the limited literature on the under-reporting by caregivers of Black children as a function of SES only allows for sparing interpretations. Future research should, therefore, utilize a more expansive definition of SES, accounting for household income, household size, neighborhood risk, and school type, as different dimensions of SES. Additionally, while Black children are more likely to experience higher rates of poverty (Gibson-Davis et al., 2020), they are also at considerable risk of experiencing achievement gaps compared to their White peers, regardless of SES (Henry et al., 2020). Thus, careful consideration in future research should be given to the inherent linkages between race and systemic barriers impacting various facets of SES outcomes (e.g., the racial wealth gap; food deserts; housing segregation; disparities in primary healthcare).

Further, while the sample in this study is inclusive of all providers who conducted neuropsychological assessment, it does not control for provider characteristics (e.g., gender, race, age, SES, number of years practicing independently). Future research should investigate the impact of provider characteristics and their ameliorative (or additive) impact on informant ratings, as well as on diagnostic decisions. Further, we do not have data on how clinicians diagnosed ADHD in the current study in terms of the assessment battery or the provider’s differential diagnoses. Accordingly, while criterion contamination may be possible, given that neuropsychologists in our department often rely on caregiver and teacher ratings in their diagnostic decision-making, it is likely that clinicians completed thorough clinical interviews and potentially performance-based testing, allowing for not only standardized testing scores but also extended behavioral observations. Specific data on neuropsychologists’ case conceptualization may provide further insight into the socioecological factors that influence neuropsychological evaluation and how this differentially affects rates of ADHD for Black and White children.

Clinical implications.

Neuropsychologists must carefully consider the weight they place on informant data in the evaluation of ADHD, particularly from teachers, as endorsements may not solely reflect teachers’ appraisal of child behavior, but may also reflect perceptions of child behavior based on demographic characteristics. As such, teacher behavioral ratings may be most useful when other data points are consistent and strongly indicative of the presence of ADHD. In other words, our findings suggest that teachers’ behavioral ratings should be used cautiously when nuanced decisions in differential diagnoses are warranted or when significant issues in conduct are endorsed and restricted to the school setting. Moreover, issues related to race should be discussed during evaluations, and neuropsychologists can take advantage of feedback sessions as an opportunity to engage caregivers of Black children in conversations on offering guidance about talking to children about racial biases, providing psychoeducation on academic and behavioral risks and parental rights in education, strengthening or repairing home-school partnerships, and connecting caregivers to resources for further reading and engagement in local advocacy groups. In order to engage in these conversations, neuropsychologists must not only deepen their knowledge of the critical issues affecting their Black patients, but should also consider the role of racein their own case conceptualization, diagnostic decisions, and unique positions as advocates, scholars, and leaders across community healthcare settings.