Recommendations for the diagnostic evaluation of ADHD include a comprehensive interview addressing the child or adolescent’s developmental, medical, and psychosocial history, including academic performance, peer relations, and family functioning (9). Other components of the evaluation regularly include symptom rating scales (e.g., Brief Rating Inventory of Executive Function [BRIEF]; 10, 11), neuropsychological testing (e.g., intelligence testing, continuous performance tasks), and a thorough physical exam to augment information collected during the clinical interview (9). Primary sources of information for younger children are parents, teachers, or other caregivers, with older children and adolescents more involved in the evaluation (12). Moreover, structured and semi-structured diagnostic interviews, including the Diagnostic Interview Schedule for Children (DISC-IV; 13), are often used to assess ADHD symptomatology, as well as the presence of comorbid psychiatric diagnoses that share similar symptoms, such as difficulty concentrating and impulsivity (e.g., Oppositional Defiant Disorder [ODD], Conduct Disorder [CD], and General Anxiety Disorders; 14). Additionally, ADHD diagnoses are often classified by one of three broad subtypes: Inattentive, Hyperactive-Impulsive, and Combination, with broad behavioral diversity within and across the categories (1).

Still, there are significant issues associated with current diagnostic processes. Comprehensive evaluations for ADHD can be time-intensive and costly, often requiring upwards of five to eight hours of testing and thousands of dollars (15). A primary care physician may diagnose to circumvent these difficulties; however, they often lack the time and expertise to differentiate ADHD from other conditions (16). Additionally, the general agreement among different informants (e.g., parents, teachers, other caregivers; 17) in different settings (e.g., school versus home; 18) has also been shown to range from fair to moderate (19, 20). Further complexity is introduced into the diagnostic process by the influence of multiple sociocultural factors that can hinder diagnostic test specificity and sensitivity and, more broadly, impact differential diagnosis. For instance, early sociocultural influences such as adverse childhood experiences and poverty have been associated with disruptive behaviors and ADHD symptoms (21). Research indicates that ethnicity may predict ADHD diagnosis and treatment type, even after controlling for socioeconomic status (22).

Such findings regarding the discrepancies in diagnostic and treatment rates of ADHD have highlighted concerns surrounding the clinical process, including the appropriateness of intelligence testing and its application among non-European/Caucasian individuals (23), and the high false-positive/negative rates of continuous performance tasks (24–26). Given the implications of the diagnosis for intervention planning, the need to substantiate ADHD diagnoses with objective clinical indicators, such as reliable physiological markers, is crucial. Furthermore, physiological measures can provide another means by which clinicians, researchers, and educators can measure treatment efficacy and develop tailored treatments to address individual physiological needs.

Historically, ADHD has been treated with potent psychostimulants such as methylphenidate or amphetamine (27, 28). While these drugs effectively manage symptoms for many youth, they are legally categorized as controlled substances and have a high likelihood of abuse (27, 29). In addition, there is limited knowledge regarding adverse side effects and long-term effects on the developing brain (30). Thus, non-stimulant medications such as antidepressants and CNS anti-hypertensives have been utilized to assist in ADHD symptom management (27–32). Unfortunately, pharmacotherapy lacks precision; neurotransmitter levels are universally altered throughout the whole brain rather than within specific, targeted regions (e.g., 33). However, observations regarding the impact of such medications have led researchers to hypothesize insufficient or reduced dopamine and norepinephrine levels as mechanisms supporting the development and maintenance of behavioral and cognitive symptoms associated with ADHD.

Psychotherapy is frequently implemented in conjunction with medication. Behavioral therapy treatments often help children navigate peer relationships, which can be impaired by difficulties with emotional self-control (34). Organizational skills training can also be integrated into treatment to target executive functioning deficits commonly exhibited by youth with ADHD (35). However, the individualistic scope and often expensive nature of psychotherapy sessions (34) frequently fail to manage the evolving and concurrent biopsychosocial influences on the developing child. In sum, neither pharmacotherapy nor psychotherapy can adequately address the causes or symptoms of ADHD. Therefore, it is necessary to continue developing a broader and more holistic view of the pathogenesis of ADHD to improve diagnosis, treatment, and quality of life for children with the disorder.

The current biomedical approach frames ADHD as a neurocognitive disorder driven primarily by dysregulated dopamine and norepinephrine levels in the brain (36, 37). This explanation oversimplifies the complexity of the human body and the series of interconnected systems that produce behavior, as illustrated in Figure 1 (38, 39). The brain prepares, plans, and instructs autonomic and voluntary activity to create complex actions such as words or movement, suggesting that top-down physiological pathways are critical factors in ADHD (40). That said, brain functioning is influenced by both external stimuli via the sensory systems and internal stimuli within the body via afferent autonomic nervous system activity and neuroendocrine pathways (41, 42). Thus, the dynamic integration between these bottom-up peripheral communication pathways and the cortical system is critical to the homeostatic modulation of the brain’s output, thoughts and behaviors (38, 39).

With that backdrop, we examined the evidence for two physiological systems implicated in cognitive and emotional dysregulation: heart rate reactivity and gut microbiota. The following research questions guide this meta-analysis:

Consistent with the Cochrane Handbook for Systematic Reviews of Interventions’ best practices on research synthesis and meta-analysis, the present study was pre-registered and accepted for transparent reporting on the National Institute for Health Research international register for prospective systematic reviews website, i.e., PROSPERO (CRD42021236819; 72).

To assess the evidence of physiological dysfunction in children with ADHD, we conducted a systematic literature search of PubMed (AP), PsycInfo (MJC), ProQuest Dissertations and Theses (TGC), and Web of Knowledge/Web of Science (JMB) databases from January 2021 to April 2021, using the following search terms for gut microbiota studies: (“attention deficit hyperactivity disorder” OR “attention deficit disorder” OR “ADD” OR “ADHD”) AND (child* OR adolescen* OR infant OR pediatric OR youth OR teen*) AND (“gut microbio*” OR “gastrointestinal microbio*” OR “gut flora” OR “dysbiosis” OR “gut brain axis”). The second outcome measure of this meta-analysis was heart rate reactivity. The following search terms were used to identify the relevant studies: (“attention deficit hyperactivity disorder” OR “attention deficit disorder” OR “ADD” OR “ADHD”) AND (child* OR adolescen* OR infant OR pediatric OR youth OR teen*) AND (“heart rate reactivity” or “heart rate responsivity” or “H.R. reactivity” or “H.R. responsivity” or “cardi* reactivity”), consistent with the guidelines of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA; 72). We examined references from retrieved empirical articles and contacted active researchers in the relevant disciplines. We also contacted researchers conducting ongoing clinical trials from publicly available technical reports to acquire additional potentially relevant data.

The criteria for inclusion were studies that (1) used case-control, cohort, or observational designs; (2) examined gut microbiota or measured average heart rate in beats per minute (bpm) during the baseline and task study phases in humans up to 19 years old with an ADHD and non-ADHD control group; (3) included relevant effect size statistics; (4) were written in English; and (5) were published between within the past decade. The exclusion criteria were (1) review articles and (2) randomized clinical trials, primarily due to a lack of a non-ADHD control group. We obtained full-text articles of all studies that met all criteria before extracting the relevant data from each article. Any article for which inclusion status was unclear during the screening process was discussed among reviewers (AP, MJC, TGC, and JMB), and its status was resolved with a consensus decision.

For all studies, ADHD was defined according to the diagnostic criteria set forth by the fourth or fifth edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV-TR or DSM-5). For details on the sample, methods, and variables of interest for all studies, please see Table 1 .

Due to the limited number of studies that met the criteria for inclusion (n =5 HRR studies; n=5 gut microbiota studies), bias control and sensitivity analyses could not be conducted. Instead, the two primary analyses included the relevant data from all eligible published gut microbiota and heart rate reactivity studies. Secondary analyses examined associations between specific bacterial phyla concentrations, heart rate reactivity task types, and symptom severity in children with ADHD compared to control groups. The Comprehensive Meta-Analysis (CMA) software was utilized to conduct all meta-analytic statistical tests (Biostat Inc., Englewood, NJ, USA). Microsoft Excel was used to calculate descriptive statistics and create a codebook for the dataset.

Literature searches yielded 395 potential studies from PubMed (n=155), PsycInfo/EBSCO (n=32), Web of Science (n=144), and ProQuest Dissertations & Theses (n=92) as shown in Figure 2 . Two additional studies were retrieved from the Reference section of included studies. After reviewing titles and abstracts for initial screening for relevant studies, 327 records were excluded. An additional 68 duplicates were removed, and 32 full-text articles were retrieved and individually assessed for eligibility. Of these 32 studies, 10 were included in the present meta-analysis; the remaining articles were ineligible because they did not provide the relevant heart rate data in bpm format (n=6), did not provide quantitative data about microbiome composition (n=11), did not have a non-ADHD healthy control group (n=1), or clarifying data were not provided following data requests for additional information (n=4).

All 10 included studies were case-control studies; the five heart rate reactivity studies and five gut microbiota studies that met the criteria for inclusion are outlined in Table 1 . In total, these studies included 531 youth with ADHD and 603 children and adolescents without ADHD, and, across the ten studies, their ages ranged from 4 to 19 years. Diagnostic methods, medication history, and comorbidities are presented in Table 2 . Only one study did not report participant medication use (74).

All studies were independently reviewed by at least two authors (AP, MJC, JMB) for quality assessment using a modified Newcastle-Ottawa Quality Assessment Scale to assess cohort and case-control studies (NOS; 77). The NOS rubric is recommended for examining the quality of case-control and cohort clinical research studies (77). Raters assessed potential bias due to selection, comparability, exposure, and reporting categories. The two authors’ scores were then averaged to obtain a single publication-quality assessment score. Only studies with scores of 8 or above out of 10 (n=10), which indicated low or no bias risk, were included in the present meta-analysis ( Table 1 ).

All five studies in the present analysis used an electrocardiogram to measure electrical heart signals (1, 14, 45, 47, 73). The type and duration of tasks are represented in Table 3 . Baseline heart rate collection ranged from 24 three-second sessions (45) to one continuous five-minute phase (47). Task-related heart rate monitoring lasted from two minutes during the emotional tasks (45, 47) to 13 minutes during a risk-taking virtual peer manipulation task (14).

The methods for assessing gut microbiology are shown in Table 4 . Across the five microbiota studies included in this analysis, the most frequently used technique for identifying, classifying, and quantifying bacteria was 16s rRNA gene amplification sequencing. The 16s rRNA gene is a well-preserved, empirically-supported, highly sensitive ribosomal fragment used to classify and quantify DNA-life forms and identify microbiota (68). Depending on the 16s rRNA gene sequence length, organisms can be identified by species, genera, families, or phyla (68).

The most frequently used indices comparing microbiome samples were the OTUS, Chao1, Shannon diversity, and Simpson measures ( Table 4 ). Each study provided different classifications and types of bacteria, which are represented in Table 4 . The OTU and alpha diversity indices of bacterial phyla were used in the present meta-analysis to minimize variation due to disparate levels of classification specificity across studies. The three commonly cited phyla included in the present meta-analysis were Actinobacteria, Bacteroidetes, and Firmicutes (60, 70).

Both fixed and random-effects models were used to assess estimation error within and between studies. Heterogeneity across studies was assessed using the I² statistic, with higher values indicating greater percentages of variation across studies that are not attributed to chance (78). A random-effects model was utilized for studies with above midpoint heterogeneity levels (I² > 50%). In all other cases, a fixed-effect model was adapted (78). The standardized mean difference (SMD) statistic was used to report on continuous variables: an SMD greater than 0 represents the children with ADHD as having greater bacterial diversity or heart rate reactivity, and an SMD less than 0 represents children with ADHD as having lower bacterial diversity and heart rate reactivity (78).

CVC, measured via task-related HRR, was assessed in the five case-control studies that met the criteria for inclusion (1, 14, 45, 47, 73). Contrary to expectations, no significant patterns emerged between the ADHD group and the controls. In addition, further analyses by task type assessing the effect size from studies that employed either a cognitive or emotional task did not find significant evidence of altered task-related HRR in the ADHD group. However, given that this research area is fairly underdeveloped, any non-significant effect sizes reported here should not be used to rule out altered autonomic reactivity among youth with ADHD.

The present study found significant between-group differences in intestinal microbial diversity across the five included studies. Specifically, the pooled small effect size of Chao1 levels in the gut profile was significantly larger for children with ADHD (60, 70, 71, 74). Follow-up analyses of the microbiota composition at the phylum level revealed that children with ADHD had a medium effect size of significantly greater concentrations of Actinobacteria. These findings support our hypothesis of altered gut microbiota in children and youth with ADHD.

Within the phylum Actinobacteria, the genus Bifidobacterium is essential during early development and is expected to taper off during childhood (60). To that end, Partty et al. (87) found that infants who later developed ADHD had Bifidobacterium deficiencies and noted the role of Bifidobacterium during the breastfeeding stages in mitigating the onset of later childhood neurodevelopmental disorders. Some researchers have observed that an increase of Bifidobacterium in the ADHD cohort occurred at the expense of more developmentally appropriate bacteria (66). Perhaps the inverse trajectory of Bifidobacterium —that is, its notable deficiency early in life and its dominance during childhood in the gut of children with ADHD— may be an important marker for assessing ADHD risk during infancy (59, 60, 70). While the specific pathway still warrants elucidation, the Actinobacteria phylum appears relevant to childhood ADHD.

Although the differences in concentrations of Bacteroidetes and Firmicutes did not reach significance, results identified trend-level differences (i.e., p <.10). The small number of studies investigating these phyla in children with ADHD may have limited the power and sensitivity of the present analysis to detect a significant small effect size. Nevertheless, control groups had greater absolute Firmicutes levels than those with ADHD in both studies reporting these bacterial concentrations (60, 70). While the differences in Firmicutes were not of a magnitude to reach traditional significance levels in each of the two studies, the shared trend of decreased Firmicutes among the ADHD group warrants future study.

In contrast, Bacteroidetes concentrations were lower in the ADHD group in the Wang et al. (70) study but higher in Aarts et al. (60). Because Bacteroidetes are connected to complex carbohydrate consumption, these mixed findings may point to the influence of diet and socioeconomic and cultural variations in diet. Future research targeting Bacteroidetes and Firmicutes composition and their ratio (88) may help address these inconsistent findings. Furthermore, assessing how a broader range of relevant sociocultural and genetic factors may influence possible physiological pathways associated with ADHD or other emotional dysregulation-related disorders may clarify the nature of these associations.

In sum, the results of this meta-analysis found evidence of altered microbiota in children with ADHD. Of note, the number of different bacteria in children with ADHD was higher than the control groups, which was an unexpected finding as greater diversity often indicates better functioning and health (76, 89). However, this perspective stems from adult populations; thus, this finding in children and adolescents requires further investigation. Second, Actinobacteria, consisting mainly of Bifidobacterium, were significantly elevated in children with ADHD. We found a medium effect size, suggesting reliable differences within this phylum could be a physiological indicator of ADHD.

Recently, differences in Actinobacteria were reported in a meta-analysis examining the gut microbiota of children with ASD, a neurodevelopmental disorder often comorbid with ADHD (75). The authors found a significant effect size for Bifidobacterium differences; however, children with ASD had lower Bifidobacterium differences such that Bifidobacterium levels were lower in children with ASD than their neurotypical peers (78). This finding contrasts with the present meta-analysis’ finding of elevated Bifidobacterium in children with ADHD (75, 78). Thus, it may be that higher levels of Bifidobacterium could serve as a sensitive indicator of ADHD. While further research is necessary to support that possibility, the nature of the present findings suggests that assessing the overall diversity of the fecal microbiota is insufficient for ascertaining ADHD-related patterns; examining the microbiota subtypes making up the gut profile is likely necessary (52).

Future research into the role of ANS activity in childhood ADHD should employ appropriate task stimuli that can best represent the circumstances and environmental contexts that contribute to externalizing symptoms in children with ADHD. Although many lab-based tasks may be stress-inducing, the nature of the stress reactivity exhibited by children with ADHD at school and home may involve different mechanisms sensitive to HRR. One such way to strengthen the external validity of study designs is to conduct studies with heart rate monitors and collect data across the day via ecological momentary assessment (EMA) methodologies (50).

As a heterogenous behavioral disorder, examining the physiological correlates of the three ADHD subtypes might provide critical information for developing tailored treatment plans. These subtypes were not consistently reported in many studies included in this report (47, 60, 71, 73, 74). Although specifying the ADHD subtype may split participants into smaller subsamples and reduce power to detect differences, categorizing may offer helpful clinical information on the differential etiology, trajectory, and treatment of ADHD.

Similarly, the findings of this meta-analysis further highlight the utility of defining ADHD biotypes. Biotyping differentiates ADHD by physiological differences irrespective of phenotypic presentation and has potential clinical and research implications (40). Different interventions have varying effects on physiological functioning; thus, identifying biotypes may enhance our ability to treat individuals with ADHD (40). For example, biofeedback therapy has been linked with decreased hyperactivity symptoms in children with ADHD (90). Specific therapies may have different effects on physiological functioning; thus, biotyping may enhance our ability to tailor ADHD treatments.

Future studies should also contribute to the growing work on differentiating ADHD by tracking information relevant to its diverse comorbidity presentation. This recommendation holds for the multiple conditions that regularly co-occur with ADHD, from other externalizing disorders (e.g., CD, ODD) and internalizing disorders (e.g., mood and anxiety disorders) to learning disabilities and other neurodevelopmental conditions (e.g., ASD; e.g., 91–94). A growing body of literature on the endophenotypes of ADHD suggests that endophenotypes may provide a mechanism for how ADHD differentially develops between siblings (95). Given that siblings may share similar diets and cultures and common genetic compositions, utilizing siblings as a control group for these studies may be an additional avenue of research.

Additionally, it will be important for researchers in this area to establish standardized protocols that permit appropriately nuanced evaluation of the microbiome and its potential contribution to ADHD. Developmental and contextual factors should be considered, given that gut microbiota composition fluctuates throughout the lifespan (40, 59, 76) and optimal gut microbiota for healthy human development are culture- and age-dependent (59). Such contextual factors could include diet composition; notably, only one study in this meta-analysis utilized a validated food frequency questionnaire (70). Finally, the current literature has not established the most efficient taxonomic level to compare gut microbiota—the present study employed two ways to assess gut microbiota: index and taxonomic unit.