Linguistic Inequities in ADHD Diagnosis among School-age Children Screened for Attention Problems in Primary Care

Abstract

Introduction.

Attention deficit/hyperactivity disorder (ADHD) remains under-identified among racial/ethnic minoritized populations. We examined whether parent reported screening questionnaires for attention problems in primary care mitigated these ADHD diagnostic inequities and identified contributing sociodemographic and clinical factors.

Methods.

We conducted a cross-sectional electronic medical record (EMR) study in an urban, hospital-based primary care pediatric clinic of school age children (N=2212) with a completed Pediatric Symptom Checklist (PSC-17). We examined differences between children with vs. without ADHD diagnoses, adjusting for positive PSC-17 attention score.

Results.

Adjusting for positive PSC attention score, children had higher odds of an ADHD diagnosis if they were English-speaking and had a documented Vanderbilt ADHD Diagnostic Rating Scale in their medical record.

Conclusion.

Multilingual, parent-report screening for attention problems in pediatric primary care does not mitigate linguistic inequities in ADHD diagnosis.

Attention deficit/hyperactivity disorder (ADHD) is one of the most common, morbid, and costly conditions in children.^1,2 If left untreated, it can lead to significant behavioral, emotional,³ social,⁴ and academic problems.^5,6 While effective, evidence-based treatments for ADHD exist that improve clinical outcomes,⁷ ADHD remains under-identified in children. This under-identification particularly affects racial and ethnic minoritized children, who suffer from an even higher ADHD prevalence and morbidity, along with higher rates of poverty, educational underachievement, and vocational failure, and yet experience disproportionate under-diagnosis and under-treatment.^8–12

Multiple studies have shown persistently higher rates of ADHD diagnosis and higher treatment utilization in non-Hispanic White children compared with other groups.^10,13–15 For example, Morgan et al. found that Black and Hispanic U.S. children had lower rates of ADHD diagnosis and treatment compared with non-Hispanic White children from kindergarten through the 8th grade.^9,10 Additionally, data from the 2011 National Survey of Children’s Health showed that children whose parents spoke English as their primary language were over four times as likely to have an ADHD diagnosis compared with children with non-English-speaking parents.^11,15 However, while these inequities are well established, there has been a lack of research demonstrating the reasons for these racial, ethnic, and linguistic disparities, nor how clinical practices can work to mitigate these disparities as part of providing care. More research is needed to help understand why these disparities exist and how they can be addressed as part of medical care delivery.

Multiple hypotheses have been proposed to explain these racial, ethnic, and linguistic disparities in ADHD diagnoses.^8,10,16–18 Most experts hypothesize that there are family-level, health care provider-level, and system-level factors that contribute to these inequities.¹⁸ Family factors include lack of problem recognition due to different cultural beliefs or values; hesitation about reporting symptoms or seeking care due to stigma or racism; competing priorities and barriers related to socioeconomic hardship; or difficulties communicating concerns to providers due to a language barrier.^{8,10,16,17,19} On the other hand, providers may harbor their own stereotypes and biases that affect their evaluation process and diagnostic decision-making.^20–22 Furthermore, providers may have difficulty communicating as thoroughly with non-English speaking families.^16,22 Finally, system level factors include structural racism, lack of culturally appropriate evaluation tools, and lack of access to interpreter services and other resources to help low English proficiency families.^16,23–25

One strategy that has been used as a tool to reduce inequities in the identification and evaluation of ADHD and other behavioral health conditions is universal and routine screening for psychiatric symptoms in pediatric primary care. Screening with standardized, parent-report questionnaires available in multiple languages—such as the Pediatric Symptom Checklist (PSC)—has been widely adopted in primary care settings to improve symptom detection. The PSC can reduce the likelihood of parent under-reporting and compensate for provider under-detection of problematic symptoms due to biases or linguistic barriers.^26,27 The PSC Attention Problem Scale has been shown to perform as well as other more extensive tools at detecting ADHD in urban settings with non-English speaking families.²⁸ However, despite this promise, it is still unknown whether multilingual universal screening, such as with the PSC, facilitates identification of attention problems and accomplishes the goal of reducing ADHD diagnostic inequities in clinical practice, or whether additional strategies are needed.

Therefore, we sought to understand whether universal screening in primary care does in fact mitigate inequities in the identification and subsequent diagnosis of ADHD. We examined sociodemographic differences in ADHD diagnosis among school-age children at an urban, safety-net hospital-based primary care practice who were screened for attention problems using the PSC in their parent’s primary language at well-child visits. We were interested in understanding whether there were sociodemographic differences in parent reporting of attention problems on universal screening, and whether variation in parent symptom reporting vs. other factors explained the sociodemographic differences found in ADHD diagnosis. Secondarily, we examined clinical factors and care processes associated with diagnostic disparities, including other diagnoses and documentation of diagnostic questionnaire results. Our ultimate goal was to determine whether screening for attention problems with parent report questionnaires in primary care would in fact mitigate the well documented ADHD diagnostic disparities, and if not, what further measures might need to be taken to eliminate inequities in the identification of ADHD.

Methods

Research design and setting.

We conducted a cross-sectional retrospective electronic medical record (EMR) study in a large primary care pediatric clinic based in an urban safety-net teaching hospital. We extracted data from the records of children ages 6–11 who arrived for a well child visit (i.e., routine health check) with any pediatric provider (including attending pediatricians, residents, and nurse practitioners) between September 1, 2016 and August 31, 2017. This dataset was extracted in preparation for a quality improvement initiative to increase detection and evaluation of attention problems in our primary care pediatric clinic, implemented the subsequent year. We included in our dataset all patients with a completed 17-item Pediatric Symptom Checklist (PSC-17) documented in their EMR during the study period. During the study period, parents completed the PSC-17 on paper in their primary language prior to meeting with their doctor, and their answers were entered electronically into an EMR flowsheet by medical assistants. The EMR flowsheet automatically scored the questionnaire, which was reviewed by primary care providers during the visit. Children who scored at risk for attention problems on the PSC-17 could be evaluated further for ADHD with the administration of Vanderbilt ADHD Diagnostic Rating Scale (VADRS) to parents and teachers. During the study period, VADRS were available to providers in the clinic to use at their discretion. We extracted from the EMR the full PSC-17 screening results, documentation of subsequent evaluation with parent and teacher VADRS, active medical and psychiatric diagnoses, and sociodemographic information including patient’s age, sex, race/ethnicity, insurance, and language.

The study was approved and designated as exempt by the Boston University Medical Campus Institutional Review Board.

Measures and variables.

Sociodemographic information.

Patient sociodemographic information including age, sex, race/ethnicity, insurance type, and language were extracted from the EMR. For this analysis, we dichotomized age into groups of older (9–11 years) vs. younger (6–8 years) school age children. As shown in Table 1, EMR variables race and ethnicity were combined and categorized accordingly by Hispanic/Latino/Spanish vs. Not Hispanic/Latino/Spanish, White, Black, Declined, and Other. For the purpose of analysis, language was dichotomized into English-speaking vs. non-English speaking. Insurance was dichotomized to commercial (including private and self-paid) vs. public.

Table 1.

SAMPLE CHARACTERISTICS OF CHILDREN BY ADHD DIAGNOSIS.

Demographic characteristics	Total Sample N = 2212	With ADHD Diagnosis: N = 264	No ADHD Diagnosis: N = 1948	Test-statistic	p-value	OR (95% CI)
Age: Mean ± SD	8.6±1.7	9.1±1.6	8.4±1.7	5.55	<.0001**
Age: n (%)
≤8 years	1064 (48.1%)	88 (33.3%)	976 (50.1%)	26.19	<.0001**	2.0 (1.5–2.6)
>8 years	1148 (51.9%)	176 (66.7%)	972 (49.9%)
Sex: n (%)
Male	1094 (49.5%)	184 (69.7%)	910 (46.7%)	49.12	<.0001**	2.6 (1.9–3.5)
Female	1118 (50.5%)	80 (30.3%)	1038 (53.3%)
Race/Ethnicity: n (%)
Hispanic/Latino/Spanish	353 (15.9%)	47 (17.8%)	306 (15.7%)	.37	.53	.7 (.3–1.2)
Non-Hispanic Black	1141 (51.6%)	155 (58.7%)	986 (50.62%)	1.1	.30	.7 (.4–1.2)
Non-Hispanic White—reference	111 (5.0%)	20 (7.6%)	91 (4.7%)	—	—	—
Non-Hispanic Declined	525 (23.7%)	36 (13.6%)	489 (25.1%)	23.25	<.0001**	.3 (.2–.6)
Non-Hispanic Other	82 (3.7%)	6 (2.3%)	76 (3.9%)	.22	.52	.6 (.4–1.3)
Language: n (%)
English	1431 (64.7%)	220 (83.3%)	1211 (62.2%)	45.6	<.0001**	3.0 (2.2–4.3)
Non-English	781 (35.3%)	44 (16.7%)	737 (37.8%)
Insurance: n (%)
Public	1643 (74.3%)	213 (80.7%)	1430 (73.4%)	6.43	.01*	1.5 (1.1–2.1)
Commercial	569 (25.7%)	51 (19.3%)	518 (26.6%)

Note:

^*p-value<.05

^**p-value<.01

ADHD=Attention-Deficit/Hyperactivity Disorder

OR = Odds Ratio CI = Confidence Interval

SD = Standard Deviation

Psychosocial screening results.

Patients were screened at well child visits with the 17-item Pediatric Symptom Checklist (PSC-17), which was provided in paper form in their primary language. The PSC-17 is a commonly used, behavioral health screening tool that has been translated into over 30 languages, including the most common languages spoken in our sample (English, Haitian Creole, Spanish, Portuguese, and French). The PSC-17 has been validated in many languages and settings²⁹ and has both high internal consistency (Cronbach’s α = 0.87) and test-retest reliability (ICC = 0.85).^29,30 Symptoms are rated on a three-point Likert scale and the items added to yield a total score, internalizing symptom score (5 items), externalizing symptom score (7 items), and attention score (5 items). Higher scores indicate higher levels of impairment in each domain.³⁰ The PSC attention score includes five items: 1) fidgety, unable to sit still; 2) daydreams too much; 3) distracted easily; 4) has trouble concentrating; and 5) acts as if driven by a motor. A positive PSC attention score (≥7) has been shown to accurately predict ADHD diagnosis at well-child visits in an urban clinic setting with majority Spanish-speaking patients, with a positive predictive value of 70.0% and a specificity of 81.3%.^28,30

At our pediatric primary care clinic, the PSC-17 is provided to families at check in for a well child visit for children aged 6–11 years old. Front desk staff assemble a packet of information and questionnaires for parents at the time of check-in in the parent’s preferred language, which is listed in the EMR and confirmed with the parent on arrival. They maintain a stack of paper PSC-17 forms and other documents in the most common languages spoken by patients seen at our hospital, including English, Spanish, Haitian-Creole, Portuguese, and French. Parents complete the PSC-17 on paper, after which the form is collected and entered into the EMR by a medical assistant, where it is autoscored and can be viewed by the provider during the visit. The provider uses the scores (attention, internalizing, externalizing, and total) to ensure detection and discussion of any behavioral or emotional concerns with the parent, and further evaluation as needed.

Psychiatric and medical diagnoses.

All psychiatric and medical diagnoses, including ADHD, were collected from each patient’s active problem list, which includes an International Statistical Classification of Diseases and Related Health Problems (ICD-10)³¹ code for each problem. Each problem is added by a medical provider to the list and remains active until a provider marks it as resolved. A patient was categorized as diagnosed with ADHD if they had any of the following ICD-10 codes on their active problem list: F90.0, F90.1, F90.2, F90.8, F90.9, F98.8, R41.840, and R46.3. ICD-10 codes for other psychiatric conditions were grouped into their corresponding diagnostic category, including autism spectrum disorders, developmental and learning disorders (including learning disabilities, language disorders, and other developmental delays), mood disorders (including depression, bipolar disorder, and disruptive mood dysregulation disorder), anxiety (including anxiety disorders, obsessive compulsive disorders, and posttraumatic disorders), and disruptive behavior disorders (including oppositional defiant disorder, conduct disorder, and unspecified behavior problems). Presence of the most common pediatric physical health conditions (obesity and asthma) were also recorded. Previous research has demonstrated the ICD-10 classification codes as a feasible method of capturing patient diagnoses.³¹

Vanderbilt ADHD Diagnostic Rating Scale.

The Vanderbilt ADHD Diagnostic Rating Scale (VADRS) is a 47-item self-report tool for parents and teachers to measure for the presence and severity of ADHD, other symptoms, and associated impairment.³² The dimensional scales of the VADRS have shown to be a reliable and broadly correlated method for detecting the presence or absence of corresponding DSM-IV- based diagnoses, and the scales are frequently used for additional assessment prior to making a diagnosis of ADHD or for tracking treatment progress.^18,32 At our clinic, the VADRS are used to further evaluate for ADHD diagnostic criteria in children who are clinically suspected to have ADHD, including children who screen positive on the PSC-17 attention scale. Paper copies of the VADRS were available throughout the study period in three languages (English, Spanish, and Portuguese) in the clinic and provided to parents to complete, as well as to give to their child’s teacher(s). The VADRS could be returned in person, by mail, or by fax to primary care providers to facilitate diagnostic evaluation of patients’ attention problems, and patient navigators entered the parent and teacher responses into EMR flowsheets to assist with scoring and documentation. For this analysis, we recorded the presence of any completed teacher or parent VADRS EMR flowsheet in a patient’s chart, and dichotomously categorized patients by any vs. no completed VADRS in the EMR.

Data analysis.

The data were analyzed using Statistical Analytical Systems (SAS) software version 9.4.³³ We used descriptive statistics to characterize the sample and then examined differences in sociodemographic and clinical characteristics between children with and without an ADHD diagnosis using t-tests, chi-squared tests, and calculating odds ratios where appropriate. For regression analyses, we used ADHD diagnosis as our primary outcome, sociodemographic characteristics—language, race/ethnicity, and insurance status—and clinical characteristics as described above as independent variables. Pediatric Symptom Checklist attention score, age, and sex were used as covariates.^29,33 We performed two multivariable logistic regression models, with and without adjusting for PSC attention score, to include sociodemographic characteristics, documented VADRS, any psychiatric diagnosis, and asthma diagnosis as main independent variables along with age and sex as covariates. For all statistical analyses, α was set to .05 for statistical significance.

Results

Study sample.

During the study period, 3,345 children ages 6–11 years old were seen for well child visits, and 2,388 (71.4%) of these children had documented PSC-17s from that visit in the EMR. Children with documented PSC-17s did not significantly differ from children without a PSC-17 based on age, gender, race/ethnicity, language, and insurance. For the analysis, we included only children with completed PSCs in their record (n=2,212), of whom 264 (11.1%) had an ADHD diagnosis. The final sample was 35.0% non-English speaking, with 18 other languages represented, most common being Haitian Creole (20.0%), Spanish (4.5%), Cape Verdean (4.2%), Tigrinya (2.3%), and Somali (1.1%).

Characteristics of children with vs. without an ADHD diagnosis.

Sample characteristics by ADHD diagnosis are presented in Table 1. Children with an ADHD diagnosis differed significantly from those without an ADHD diagnosis based on age, sex, language, insurance status, and race/ethnicity. Children had significantly higher odds of an ADHD diagnosis if they were older (age 9–11 vs. 6–8 years, X²= 26.19; p-value <.0001; odds ratio [OR] 2.0; 95% CI [1.5–2.6]), male (X²= 49.12; p-value <.0001; OR 2.0; 95% CI [1.9–3.5]), had English listed as the primary language (English vs. non-English, X²= 45.6; p-value <.0001; OR 3.0; 95% CI [2.2–4.3]), and had public insurance (X²= 6.43; p-value =.01; OR 2.0; 95% CI [1.5–2.6]). Children whose race was reported as “declined” had significantly lower odds of an ADHD diagnosis compared with children identified as non-Hispanic White (X²=23.25; p-value<.0001; OR 0.3; 95% CI [0.2–0.6]).

Children with a diagnosis of ADHD had higher psychiatric and medical complexity, which is presented in Table 2. Diagnosis of any psychiatric disorder other than ADHD (including disruptive behavior, anxiety, mood, autism spectrum, and other developmental disorders) was significantly higher among children with an ADHD diagnosis than those without an ADHD diagnosis (58.3% vs. 21.6%; X²= 162.9; p-value = <.0001; OR 5.1; 95% CI [3.9–6.6]). Children with an asthma diagnosis also had significantly higher odds of diagnosed ADHD (X²=46, p-value <.0001, OR 2.6; CI [1.9–3.5]), but not children with an obesity diagnosis. Children had significantly higher odds of an ADHD diagnosis if they had a completed parent or teacher VADRS in the EMR (17.4% vs. 2.8%; X²= 113.7; p-value < .0001; OR 7.3; 95% CI [4.8–11.0]).

Table 2.

CLINICAL CHARACTERISTICS OF CHILDREN WITH VS. WITHOUT AN ADHD DIAGNOSIS

Clinical characteristics	Total Sample N = 2212	With ADHD Diagnosis: N = 264	No ADHD Diagnosis: N = 1948	Test-statistic	p-value	OR (95% CI)
Diagnosis: n (%)
Non-ADHD psychiatric diagnosis
Anxiety	575 (26%)	154 (58.3%)	421 (21.6%)	162.9	<.0001**	5.1 (3.9–6.6)
Autism	82 (3.7%)	25 (9.5%)	57 (2.9%)	27.8	<.0001**	3.5 (2.1–5.7)
Disruptive behavior	59 (2.8%)	16 (6.1%)	43 (2.21%)	13.3	.0003**	2.9 (1.6–5.2)
Developmental problems	161 (7.3%)	71 (26.9%)	90 (4.6%)	170.9	<.0001**	7.6 (5.4–10.7)
Mood disorder	415 (18.8%)	102 (38.6%)	313 (16.1%)	77.7	<.0001**	3.2 (2.5–4.3)
Medical comorbidities	37 (1.8%)	8 (3.0%)	29 (1.5%)	3.36	.06	2.1 (.9–4.6)
Asthma	384 (17.4%)	85 (32.2%)	299 (15.4%)	46.0	<.0001**	2.6 (1.9–3.5)
Obesity	633 (28.6%)	80 (30.3%)	553 (28.4%)	.42	.5	1.1 (.8–1.5)
Documented Vanderbilta: n (%)	101 (4.6%)	46 (17.4%)	55 (2.8%)	113.7	<.0001**	7.3 (4.8–11.0)
Positive PSC-17b scores: n (%)
Internalizing (≥ 5)	165 (7.5%)	49 (18.6%)	116 (5.9%)	53.52	<.0001**	3.6 (2.5–5.2)
Externalizing (≥7)	206 (9.3%)	61 (23.1%)	145 (7.44%)	67.53	<.0001**	3.7 (2.8–5.2)
Attention (≥7)	253 (11.4%)	109 (41.3%)	144 (7.4%)	263.7	<.0001**	8.8 (6.5–11.9)
Total (≥15)	255 (11.5%)	90 (34.1%)	165 (8.5%)	149.6	<.0001**	5.7 (4.1–7.5)
PSC-17 scores: Mean ± SD
Internalizing	2.6±2.6	2.4 ±2.3	1.4 ±1.7	8.35	<.0001**
Externalizing	5.5±3.2	4.3 ±2.9	2.3 ±2.4	11.9	<.0001**
Attention	3.3±2.5	5.9 ±2.3	2.9 ±2.3	19.7	<.0001**
Total	7.4±5.6	12.6.2±5.9	6.7±5.3	15.5	<.0001**

Note:

^*p-value<.05

^**p-value<.01

ADHD=Attention-Deficit/Hyperactivity Disorder

OR = Odds Ratio CI = Confidence Interval

^aVanderbilt = Vanderbilt ADHD Diagnostic Rating Scale

^bPSC-17 = Pediatric Symptom Checklist, 17 item version

SD=Standard Deviation

Children with an ADHD diagnosis had a significantly higher mean PSC-17 internalizing subscale score (2.4 vs. 1.4; t-value = 8.35; p-value < .0001), externalizing subscale score (4.3 vs. 2.3; t-value = 11.9; p-value < .0001), attention subscale score (5.9 vs. 2.9; t-value = 19.7; p-value < .0001), and total score (12.6 vs. 6.7; t-value = 15.5; p-value < .0001) compared with children without an ADHD diagnosis.

Impact of attention screening on ADHD diagnostic disparities.

To understand the impact of positive screens for attention problems on ADHD diagnostic disparities in our sample, we performed multivariable logistic regression analyses both with and without adjusting for positive PSC attention score (Table 3), controlling for age and sex. Both models yielded comparable results. In the unadjusted model, children had significantly higher odds of an ADHD diagnosis if they were English-speaking (Wald X² = 24.9; aOR 2.6; 95% CI [1.7–3.8]; p-value <.001), had a documented VADRS in the EMR (Wald X² = 44.4; aOR 4.8; 95% CI [3.0–7.7]; p-value <.0001), had another non-ADHD psychiatric diagnosis (Wald X² = 75.1; aOR 3.6; 95% CI [2.7–4.7]; p-value <.0001), or had an asthma diagnosis (Wald X² = 12.2; aOR 1.8; 95% CI [1.3–2.4]; p-value=.005). In the model adjusted for PSC attention score, children also had significantly higher odds of an ADHD diagnosis if they were English-speaking (Wald X² = 16.5; aOR 2.2; 95% CI [1.5–3.3]; p-value <.001) had a documented VADRS in their EMR (Wald X² = 26.8; aOR 3.6; 95% CI [2.2–5.9]; p-value <.0001), had a non-ADHD psychiatric diagnosis (Wald X² = 44.9; aOR 2.8; 95% CI [2.1–3.8]; p-value <.001), or had an asthma diagnosis (Wald X² = 7.4; aOR 1.6; 95% CI [1.1–2.2]; p-value=.0063).

Table 3.

ASSOCIATION BETWEEN ADHD DIAGNOSIS AND DEMOGRAPHIC CHARACTERISTICS AMONG SCHOOL-AGE CHILDREN

		aOR (95% CI)
Outcome	Sociodemographic characteristics	Sociodemographic adjusted	Sociodemographic + attention score adjusted
ADHD Diagnosis	Age: ≤8 vs. >8 years	1.9 (1.4–2.6)*	2.4 (1.7–3.3)*
	Sex: male vs. female	2.1 (1.6–2.8)*	2.0 (1.5–2.8)*
	Ethnicity:
	Hispanic vs. non-Hispanic White	.7 (.4–1.5)	.7 (.3–1.4)
	Non-Hispanic Black vs. non-Hispanic White	.8 (.5–1.5)	.9 (.5–1.6)
	Other vs. non-Hispanic White	.6 (.3–1.3)	.7 (.3–1.3)
	Insurance: public vs. commercial	1.4 (.9–2.0)	1.3 (.9–1.9)
	Language: English vs. non-English	2.6 (1.8–3.8)*	2.2 (1.5–3.3)*
	Documented Vanderbilta: Yes vs. No	4.8 (3.0–7.7)*	3.6 (2.2–5.9)*
	Any Psych Diagnosis: Yes vs. No	3.6 (2.7–4.7)*	2.8 (2.1–3.8)*
	Asthma Diagnosis: Yes vs. No	1.8 (1.3–2.4)*	1.6 (1.1–2.2)*

Note:

^*p-value<.05

ADHD=Attention-Deficit/Hyperactivity Disorder

aOR = Adjusted Odds Ratio and CI = Confidence Interval

^aVanderbilt = Vanderbilt ADHD Diagnostic Rating Scale

Discussion

In a cohort of urban school-age children who were universally screened for attention problems, we found that children from non-English-speaking families were less likely to be diagnosed with ADHD than children from English-speaking families, even when accounting for parent-reported attention problems in primary care. Thus, in this population, the findings suggest that the disparity in ADHD diagnosis related to a parent’s primary language was not primarily explained by parental symptom disclosure. Even when non-English speaking parents report high levels of attention problems on a pediatric screening tool in their native language, their children still have significantly lower odds of being diagnosed with ADHD and getting the help they need. To our knowledge, this is the first study to examine factors associated with ADHD diagnosis among school-age children screened for attention problems in a primary care setting.

Our finding that a non-English primary language reduces the likelihood of an ADHD diagnosis extends previous literature on linguistic inequities in ADHD diagnosis. In the 2011 National Survey of Children’s Health, children living in a household where English was the primary language were four times more likely to be diagnosed with ADHD than those in which English was not the primary language.¹¹ Previous literature has found that linguistic barriers are associated with a substantial reduction in both access to and quality of health care,²⁵ and specifically can interfere with accurate parent reporting of ADHD symptoms to their care provider.³⁴ In our study, while non-English speaking parents did report ADHD symptoms on the PSC-17, children were still not diagnosed at rates similar to children of English-speaking parents. It could be that parents reported fewer concerns in follow-up clinical interviews or refused further evaluation of ADHD due to cultural and community beliefs,³⁵ or fear of stigmatization.^36,37 However, our findings also suggest that the linguistic barrier likely also interferes with the subsequent diagnostic evaluation, including obtaining follow-up diagnostic questionnaires from both parents and teachers, which are typically used to make the diagnosis of ADHD in primary care. In fact, we found that having the Vanderbilt ADHD Diagnostic Rating Scale (VADRS) from a parent or teacher documented in the EMR was associated with some of the highest odds of an ADHD diagnosis in this population. This was not surprising given the intended use of these questionnaires but may highlight one of the most important gaps in care for non-English speaking families. Some non-English speaking families may not be able to complete diagnostic questionnaires due to lack of availability in their language, given that the VADRS was only available during this period in three languages (English, Spanish, and Portuguese), which does not include the most common primary language after English in our sample (Haitian Creole). The linguistic barrier may also interfere with the process of coordinating with teachers to obtain questionnaire results, highlighting the complexity of obtaining these measures from both parents and teachers. Finally, and importantly, our study suggests that provider bias is an important consideration, which may cause inadvertent dismissal of ADHD concerns in families whose primary language is not English. Providers may be more inclined to attribute symptoms and school concerns to cultural differences or linguistic barriers.

Children whose race was listed as declined had significantly lower odds of an ADHD diagnosis, but this difference was not statistically significant in the regression model. This finding is not consistent with previous studies that have shown significantly lower rates of ADHD diagnosis in African American and Hispanic children compared with non-Hispanic White children.^{2,9,10,38–40} One explanation is that we may not have been well-powered to detect a racial or ethnic disparity in ADHD diagnosis due to the low percentage of non-Hispanic White patients in our sample, and the high sample of individuals who declined to disclose their race. However, our results also suggest that universal screening may improve some underdiagnosis due to racial and/or ethnic biases, and that some of this disparity is related to primary language.

Children with psychiatric comorbid diagnoses also had significantly higher odds of an ADHD diagnosis than those without comorbidities. This aligns with the fact that ADHD often presents with a wide range of mental health and neurodevelopmental comorbidities^41–43 that share common environmental and genetic risk factors with ADHD⁴⁴ and may be a direct reflection of the impact of ADHD symptoms. For example, poor academic achievement and social difficulties linked with ADHD may lead to the onset of anxiety disorders, depression, and poor psychosocial functioning. This finding also may reflect the fact that comorbid disorders are associated with worse symptoms and functional impairment in children with ADHD leading to a higher likelihood of diagnosis in part due to comorbidities. In addition, it may be that medical or psychiatric attention for one disorder facilitates diagnosis of another, whereby comorbid presentations may highlight ADHD indicators and lead parents or teachers to request an evaluation.

Finally, our study showed that children had higher odds of an ADHD diagnosis if they were also diagnosed with asthma. Many previous studies have shown a relationship between ADHD and asthma,⁴⁵ as there may be both genetic and environmental influences contributing to this high comorbidity. This finding suggests that the presence of an asthma comorbidity might be a facilitator to diagnosis of ADHD in affected children. Similar to psychiatric comorbidity, this may be because their functional impairment from both disorders combined is more severe, or because children with asthma may be more connected to medical providers and/or have better access to health care services.

The findings from this study should be considered in light of its limitations. First, as this was a retrospective chart review of cross-sectional data, we cannot establish causality of the associations found. Secondly, this was not a longitudinal study, and we did not know the length of time between ADHD diagnoses and positive attention screens. Third, children in our study were mostly publicly insured and non-White; thus, our findings may not generalize to other geographic and clinical populations. Finally, due to the sample size, we may have been inadequately powered to detect other clinically meaningful differences.

There are several key clinical and practice implications from study findings. Research has demonstrated that perceived discrimination from providers is linked with poorer quality and underutilization of mental health services among racial/ethnic minoritized and non-English speaking families.^46–48 In particular, Hispanic patients report greater levels of distrust toward mental health professionals,^37,48 poorer provider-patient alliance,⁴⁹ more experiences of provider dismissiveness and invalidation,^20,49 and substandard care.⁵⁰ This inequity may be attributed to health care providers’ biases about their racial/ethnic minoritized patients during diagnostic assessment.^21,50 During a clinical visit, a provider may unconsciously make insensitive remarks about a patient’s racial or cultural background, contributing to self-doubt, frustration, distrust of providers and systems, and lower treatment utilization.⁵¹ Furthermore, the presence of such biases may contribute to providers not conducting clinical follow-up and diagnostic questionnaires (i.e. VADRS), leaving out important pieces of the diagnostic picture.^22,51 These treatment disparities are further perpetuated by lack of language interpreters,^10,51 greater presence and burden of stigma,^52,53 and historical racial maltreatment in health care settings.⁵⁴ Multicultural competency training interventions and expanding organizational support of training provide strong potential avenues to reduce these biases and train health care providers with the knowledge, skills, and sensitivity to respond to sociocultural issues within clinical and diagnostic settings.^55–59 Training should focus on developing provider’s awareness of their own influences, prejudices, and biases, cross-cultural communication skills for Non-English speaking patients, and assessment of the impact of identity and marginalization on symptom presentation and patient-provider interactions.⁶⁰ Furthermore, continuous efforts should be made to expand diversity of health care providers to alleviate potential issues of cross-cultural biases and mistrust.

Finally, considering the study finding that children with ADHD had higher odds of diagnosis with a completed parent or teacher VADRS and that some non-English speaking families may not be able to complete the scale due lack of language availability, the field urgently needs to adapt and validate existing diagnostic tools in other languages. Because the diagnosis of ADHD relies on collecting assessments from multiple individuals and settings including parents and teachers,^61–63 strategies to improve school-health care-home partnerships are critical and should be developed and adapted specifically for non-English speaking families to avoid diagnostic delay. For example, clinics can develop standardized provider decision-making algorithms and clinical operation workflows to streamline multi-lingual PSC-17 and VADRS distribution to parents and teachers.

In response to these findings, we initiated a multidisciplinary quality improvement initiative to increase detection and evaluation of attention problems in our primary care pediatric clinic. First, we simplified PSC-17 administration by integrating multiple languages onto a single form to increase the percentage of parent completions during clinical visits. Second, we added a new EMR feature to facilitate provider detection and recognition of positive and elevated PSC-17 attention scores in need of further evaluation. Third, we created an explanatory packet to accompany the VADRS to aid return rate, including a cover letter with pediatric clinic contact information, questionnaire information, and a release of information (ROI) for the school to facilitate communication between the school and health care providers. Finally, we pilot tested a new electronic collection of parent and teacher VADRS in four languages (English, Spanish, Portuguese, and Haitian Creole) to facilitate collection and entry of scale into the EMR.

Conclusion.

Our findings highlight that even among children whose parents report inattention and hyperactivity on a standard pediatric screening tool, significant disparities exist in the progression to clinical diagnosis. Non-English-speaking families are particularly vulnerable to diagnostic inequities, and screening in parents’ primary language is not enough to solve this problem. Efforts to couple screening with a standardized and algorithmic process for diagnostic evaluation for children who screen positive for attention problems in primary care could improve linguistic and other inequities in ADHD diagnosis for urban children.