Introduction

Mindfulness is the deliberate awareness of the present moment without judgment (Kabat-Zinn 2003). Mindfulness meditations involve selecting a point of focus, such as the breath, or a physical action such as raising and lowering arms, and regulating and directing attention to that point with sustained focused attention (Bishop et al. 2004). If the mind wanders from the point of focus, then mindfulness involves acknowledging the thought or feeling that arises, inhibiting rumination, and switching attention back to the point of focus (Bishop et al. 2004). When mindfulness is the foundation of a movement-based practice, such as Hatha Yoga, and emphasis is placed on interoceptive, proprioceptive, and kinesthetic aspects of the experience in addition to the mindfulness aspects, this kind of mindful movement practices encourages an embodied experience of the self as well (Schmalzl et al. 2014).

Several investigators have proposed theoretical accounts of how the practice of mindfulness can enhance and develop attention regulation (Bishop et al. 2004; Lutz et al. 2008). They proposed four types of attention regulation involved in mindfulness: sustained attention on the present moment (i.e., by focusing on a target object), monitoring the present moment (i.e., detect mind wandering), executive function abilities such as attentional switching (i.e., disengage from a distracting object/thought without further involvement), and selective attention (i.e., ability to redirect focus promptly back to the target object).

Attention and executive function (EF) underlies most behavior from childhood onward (Douglas 1972; Tannock and Schachar 1996). Attention is a cognitive ability that regulates the amount of information we take in and acts as a “spotlight” (Cohen 2014). It is related to most cognitive and neuropsychological functions in our everyday life (Cohen 2014), such as EF processes. EF is an umbrella term for cognitive processes such as self-control (inhibition), decision-making, goal setting, planning, problem solving, emotional responses, and behavior (Lezak 2012).

Attention and EF is required to perform everyday activities. Deficits in attention or EF are likely to influence a child’s behavior, self-regulation, and academic abilities (Carver and Scheier 2012). Such disruption in attention is often associated with behavioral characteristics of childhood neurodevelopmental disorders including cerebral palsy (Bax et al. 2005), attention-deficit hyperactivity disorder (ADHD) (Brocki and Bohlin 2006), autism spectrum disorders (Joseph et al. 2005), and behavioral problems such as bullying and delinquency (Hughes et al. 2000). The growing body of research demonstrating concurrent and longitudinal associations between deficits in attention, socio-emotional development, and academic performance is indicative of the importance of attention and its impact across different areas of development (Blair and Razza 2007; Hughes et al. 2001; Riggs et al. 2004).

Chiesa et al. (2011) conducted a systematic review of the efficacy of mindfulness training on cognitive abilities in adults. The review identified 23 studies, including 15 RCTs or controlled trials (CT) and 8 case-control studies. Their results provide preliminary evidence that mindfulness-based practices can enhance attention and working memory capacity, although limitations in the quality of the existing research are noted. Systematic literature reviews with the pediatric literature have noted that yoga is a promising intervention for physical rehabilitation (Galantino et al. 2008); physical fitness; cardiorespiratory effects; motor skills/strength; mental health and psychological disorders, behaviors, and development; and irritable bowel syndrome (Birdee et al. 2009). Yet, a systematic literature review focusing on attention has not been conducted in the pediatric population.

The first aim is to review the current literature on mindfulness-based interventions for attention and EF in children and adolescents. The second aim is to examine mindfulness outcomes within the included studies.

Method

Search Strategy

The following databases were searched: PubMed, PsycINFO, CINAHL, Web of Science, and Scopus from 1972 to 2016 were comprehensively searched. The search strategy comprised the following MeSH headings or Keywords: Yoga OR mindfulness OR “mindful awareness” OR meditation AND; Child OR children OR adolescence OR adolescent OR paediatric OR pediatric AND; Cognition OR attention OR cognitive function OR executive function. Studies were downloaded into Endnote 15, and duplicates were deleted. Studies were identified by title and abstract and screened by the authors to assess whether they met the selection criteria set out below. The reference lists of relevant systematic reviews were screened for additional references, with snowballing used to ensure that all relevant papers were identified.

Inclusion Criteria

Studies were included in this systematic review if they were randomized controlled trials and quasi-randomized controlled trials (e.g., randomization by group); interventions with a focus on yoga, meditation, and/or mindfulness-based techniques; yoga interventions which incorporated asana (yoga postures) or pranayama (yogic breathing) or yoga nidra (yogic relaxation) and/or meditation for the children or adolescents; mindfulness-based interventions which included mindfulness meditations and/or other mindfulness exercises, such as mindful eating, mindful walking, or Tai Chi; and if study participants were children/adolescents aged between 5 and 18 years old. Studies were required to have attention or executive function as an outcome measure; yoga and/or mindfulness interventions that incorporated other modalities, such as interactive discourse and non-specified relaxation techniques, were included; and dissertations were included.

Consequently, studies were excluded if they did not provide at least one adequate measure of child attention or executive function outcome, such as if study only measured overall child ADHD symptoms without examining attention specifically. Papers outside the peer-reviewed literature that were not dissertations were also excluded.

The full search yield was initially reviewed for inclusion by two independent reviewers (first and second author) on the basis of title and abstract. We contacted the first authors of two non-English papers (Bueno and Delgado 2015; Haffner et al. 2006) to ask if they have an available English translation. One author (Haffner et al. 2006) sent an English-translated version of their paper, and the other author (Bueno and Delgado 2015) offered to translate their Spanish paper into English; however, after a brief discussion, it became clear that the paper did not fit our inclusion criteria. Both reviewers then assessed the full text of the remaining articles for adherence to the inclusion criteria, and discrepancies were resolved by discussion. We also contacted the first authors of four of the included studies for additional data and details of their studies. One author replied with the additional information required (Semple et al. 2010), while three authors unfortunately could not be contacted (Kratter 1983; Leonard et al. 2013; Verma et al. 1982).

Methodological Quality Assessment

Methodological quality of included studies was assessed using the Physiotherapy Evidence Database (PEDro) Scale. The studies were assessed by first and second authors independently. Discrepancies were resolved through discussion.

Assessment of Risk of Bias

The risk of bias of the included studies was assessed using the Cochrane Collaboration’s tool (Higgins and Green 2008). The tool looks at the six domains of bias. These domains include selection bias (whether sequence adequately generated and allocation adequately concealed prior to assignment), performance bias (was there blinding of participants and personnel or was the knowledge of group/intervention allocation by participants and personnel during the study adequately prevented?), detection bias (was there blinding of outcome assessors or was the knowledge of group/intervention allocation by outcome assessors adequately prevented?), attrition bias (were amount and nature of handling of incomplete outcome data adequately addressed?), reporting bias (was report of the study free of suggestion of selective outcome reporting?), and other bias (was the study free of other problems not covered above that could put it at a high risk of bias?).

First and second authors independently made judgments for each of the domains but selecting “low risk,” “high risk,” or “unclear risk” of bias. Unclear risk of bias was selected when insufficient detail was reported or the risk of bias was unknown. Discrepancies were resolved through discussion. The risk of bias of the included studies is presented in the risk of bias graph (Fig. 1) and risk of bias table (Fig. 2).

Fig. 1
figure 1

Risk of bias graph across included studies

Full size image
Fig. 2
figure 2

Risk of bias summary for all included studies

Full size image

Data Extraction

Data extracted from each study included study design, participant characteristics, intervention characteristics, and the attention and/or executive outcome measures and mindfulness measures. The first author extracted and tabulated the relevant data from the studies, and any queries were clarified with the second author.

Data Synthesis

Relevant quantitative outcome data from each study were analyzed to determine a measure of intervention effect size. The reported means, standard deviations, and sample size for control and treatment groups at post-intervention time point were used to calculate a t test value using Hedges’ g, as illustrated in the equation below (mean difference/pooled standard deviation), and to determine if there was a significant difference between the groups after the intervention. Based on the guidelines suggested by Cohen (1992), effect sizes were classified as small (0.2), medium (0.5), or large (0.8).

$$ \frac{M_{\mathrm{postT}}-{M}_{\mathrm{postC}}}{\sqrt{\left({SD}_{{\mathrm{posT}}^2}\left({n}_T-1\right)+{SD}_{{\mathrm{postC}}^2}\left({n}_C-1\right)\right)/\left({n}_T+{n}_C-2\right)}} $$

The authors intended to conduct a meta-analysis on the collated outcome data using RevMan 5.0. Due to the substantial variation in the population, the measures used, and the outcomes assessed in the included studies, however, only a small meta-analysis with two studies were conducted.

Results

Descriptions of Studies

A total of 1034 articles were identified from the databases (see Fig. 3) using the search strategy described above. Two additional references found from the identified systematic reviews were included. Three hundred sixty-one duplicated articles were excluded, and 673 articles were screened by title and abstract. Of these, 651 articles were excluded, as they clearly did not meet the inclusion criteria. A total of 22 papers were retrieved to consider in further detail, of which 13 met inclusion criteria.

Fig. 3
figure 3

Flow chart of article screening—included and excluded studies

Full size image

Settings

Eight of the 13 studies included in this review were carried out in the USA (Britton et al. 2014; Felver et al. 2014; Flook et al. 2010; Kratter 1983; Leonard et al. 2013; Moretti-Altuna 1987; Semple et al. 2010; Sidhu 2013). Five of these studies were part of a dissertation (Felver et al. 2014; Kratter 1983; Moretti-Altuna 1987; Semple et al. 2010; Sidhu 2013). Two studies were carried out in India (Telles et al. 2013; Verma et al. 1982), one in Germany (Haffner et al. 2006), and one in Canada (Schonert-Reichl et al. 2015).

Nine of the 13 included studies were published between 2010 and 2016 (Britton et al. 2014; Felver et al. 2014; Flook et al. 2010; Leonard et al. 2013; Purohit and Pradhan 2016; Schonert-Reichl et al. 2015; Semple et al. 2010; Sidhu 2013; Telles et al. 2013), while one study was published in 2006 (Haffner et al. 2006), and three studies were published between 1982 and 1987 (Kratter 1983; Moretti-Altuna 1987; Verma et al. 1982).

Participants

As detailed in Table 1, the age and population varied across the studies. For the purpose of this review, only the data for the non-treatment control group and the treatment mindfulness-based intervention group of the included studies were included. The data for other available comparison groups such as medication (Moretti-Altuna 1987) or relaxation (Kratter 1983) comparison groups were not included.

Table 1 Sample characteristics and experimental design of included studies
Full size table

Ten studies consisted of children (ranged between 7 and 12 years). Five of these studies recruited typically developing children (Britton et al. 2014; Felver et al. 2014; Flook et al. 2010; Schonert-Reichl et al. 2015; Telles et al. 2013), while four of the studies recruited children with ADHD (Haffner et al. 2006; Kratter 1983; Moretti-Altuna 1987; Sidhu 2013) and one recruited children with reading difficulties (Semple et al. 2010). Three studies recruited adolescents (ranged between 11 and 18 years), and of these three studies, one recruited adolescents in correctional schools (Verma et al. 1982), another study recruited incarcerated adolescents (Leonard et al. 2013), and the third study recruited adolescent orphans (Purohit and Pradhan 2016).

The first author for one of the included studies could not be contacted when more information on their study was requested (Kratter 1983). As a consequence, the missing information, number of participants per group, was listed as unknown.

Types of Intervention

Three categories of mindfulness-based interventions were identified: yoga intervention (Haffner et al. 2006; Purohit and Pradhan 2016; Telles et al. 2013), mindfulness-based psychological interventions (Britton et al. 2014; Felver et al. 2014; Flook et al. 2010; Leonard et al. 2013; Schonert-Reichl et al. 2015; Semple et al. 2010; Sidhu 2013), and traditional meditation training (Kratter 1983; Moretti-Altuna 1987; Verma et al. 1982) (see Table 1).

All interventions included a component of body-awareness training (i.e., observing the breath), although in some interventions, this was not the primary focus. Interventions were delivered in a variety of ways across the studies, but not all the studies reported how the interventions were delivered or who delivered them. Of those that reported this information, two studies reported that the intervention was delivered by people trained specifically in the study’s intervention techniques (Felver et al. 2014; Telles et al. 2013), one delivered by clinicians trained in mindfulness training and cognitive behavior therapy (Leonard et al. 2013), and another by researchers (Sidhu 2013), while one reported that the superintendent of the correctional centers delivered the intervention (Verma et al. 1982), and two studies reported that school teachers delivered the intervention (Britton et al. 2014; Schonert-Reichl et al. 2015). Two of the 13 studies invited parents to participate in the mindfulness intervention with their children; one of these studies allowed the parent to join their children after two parent-only sessions (Semple et al. 2010) and the other study invited the parents to attend the sessions with the children straight away, but each session consisted of a period of time where parents and children participated separately (Felver et al. 2014).

The duration, intensity, and dosage of the interventions varied across the 13 studies. Duration of interventions ranged widely from 3 to 24 weeks, while the dosage also ranged widely from 135 to 4320 min. It should be noted that the dosage of one of the studies could not be calculated because the duration of each session was unknown (Verma et al. 1982).

Measures

A range of measures were used across the studies to measure attention and EF as detailed in Table 2. Some of the measures were self-report questionnaires from the perspective of the child/adolescent, parent, or teacher, while other measures were objective neuropsychological assessments such as computer-orientated tasks or pen-paper task that assess particular aspects of attention or EF. Within the 13 studies, 21 different measures of attention and EF were used (see Appendix for a description of the measures by the included studies). The Attention Network Task (ANT), the Stroop Test, and the Fruit Distraction Test (FDT) were the only attention and EF measures used more than once across the 13 studies. The ANT is a computer task used to measure attention, and it was used in two studies (Felver et al. 2014; Leonard et al. 2013); however, a meta-analysis could not be conducted on the data from these two studies because the data could not be pooled together. This is because one of the studies reported a mean and standard deviation for each of the ANT subsystems (Felver et al. 2014), while the other study reported an overall mean and standard error across the three subsystems (Leonard et al. 2013). In addition, the population of the two studies were too diverse; one population consisted of healthy children between 9 and 12 years (Felver et al. 2014), and the other population consisted of incarcerated youths between 16 and 18 years old (Leonard et al. 2013). The Stroop Test, a color-word naming task used to measure EF, was also used in two studies (Purohit and Pradhan 2016; Telles et al. 2013), and a small meta-analysis was conducted using the pooled Stroop Test data from these two studies. The findings of the meta-analyses are reported under the “Findings of Studies” section of this paper. Finally, the FDT was another EF test that was used in two studies (Kratter 1983; Moretti-Altuna 1987). A meta-analysis could not be conducted, however, because the number of participants in each group in the analysis was not reported in one of the studies (Kratter 1983).

Table 2 Attention and executive function measures used in the studies reviewed
Full size table

Computerized Measures

Apart from the ANT mentioned above, there were three other different computer tasks used by two studies to measure attention and EF in children and adolescents. One study (Sidhu 2013) used the Test of Variables of Attention (TOVA), a continuous performance task used to measure attention, while another study (Schonert-Reichl et al. 2015) used the computerized Flanker task and the Hearts and Flowers task to measure EF.

Pen-Paper Assessment Measures

As mentioned before, two studies used the Stroop test and two other studies used the Fruit Distraction Test to measure EF. Four other paper-and-pencil assessment tasks were used to measure attention and EF. These tasks were the Dortmund Attention Test (Haffner et al. 2006), Cancellation of number 9 (Verma et al. 1982), the Trail Making Test (Purohit and Pradhan 2016), and the Digit Symbol Substitution Test (Purohit and Pradhan 2016).

Self-Report or Parent/Teacher Report Questionnaires

Six of the 13 included studies used questionnaires to measured attention or EF outcomes. They were either measured directly by the questionnaires or the questionnaires consisted of an attention or EF subscale. One of these six studies used the Youth Self-Report Scale with an attention problem subscale (Britton et al. 2014); another study used the Teacher and Parent report versions of the Behavior Rating Inventory of Executive Function (BRIEF) (Flook et al. 2010). Parent versions of the Child Behavior Checklist—Attention Problem Subscale (CBCL) (Semple et al. 2010) and the FBB-HKS (Fremdbeurteilungsbogen für Hyperkinetische Störungen; Brühl et al. 2000), a rating scale for ADHD symptoms (Haffner et al. 2006), were used to measure attention problems or deficits in children. One study used an analog scale of the teachers’ rating of children’s attention (Telles et al. 2013). Finally, one study used both the Cognitive Problems/Inattention subscale of the Conners’ Parent Rating Scale-Revised: Long (CPRS-R:L) and the Attention Problem Scale of the Behavior Assessment System for Children (BASC-2) to measure outcome of attention (Sidhu 2013).

Although some studies reported additional variables such as clinical symptom conditions such as depression, anxiety, conduct disorder, and oppositional defiant disorder (Britton et al. 2014; Semple et al. 2010), emotional and child behavior (Britton et al. 2014; Semple et al. 2010; Sidhu 2013), self-esteem (Telles et al. 2013), child’s impulsivity and hyperactivity (Kratter 1983; Moretti-Altuna 1987), overall ADHD symptoms (Kratter 1983; Moretti-Altuna 1987; Sidhu 2013), physical activity levels (Kratter 1983; Moretti-Altuna 1987; Telles et al. 2013), other cognitive function (Moretti-Altuna 1987; Verma et al. 1982), academic performance (Schonert-Reichl et al. 2015; Telles et al. 2013), physiological measurements (Schonert-Reichl et al. 2015), and general well-being (Schonert-Reichl et al. 2015), these were considered beyond the scope of this review and therefore were not reported.

Quality Assessment

Thirteen studies scored between 4 and 9 points out of a total of 11 points on the PEDro Scale (Table 3), with an average score of 6.62. This suggests that the included studies consisted of moderate methodological quality. There were two of the 11 potential points for methodological strengths that were impractical to obtain in an RCT of a mindfulness-based intervention, where only one study reported the masking of participants to treatment (Schonert-Reichl et al. 2015) and only one study reported masking of therapists to treatment allocation (Purohit and Pradhan 2016).

Table 3 Methodological quality assessment of included studies—PEDro Scale
Full size table

Findings of Studies

Data for child and adolescent attention or EF outcomes for each study were tabulated in Table 4. The effect sizes reported in Table 4 are those calculated specifically for this review. Based on this review’s effect size calculations, five of the 13 studies found at least one significant intervention effect for attention or EF with medium to large effect sizes (0.3–32.03) (Felver et al. 2014; Haffner et al. 2006; Leonard et al. 2013; Purohit and Pradhan 2016; Sidhu 2013). Eight studies did not find significant intervention effects (Bogels et al. 2008; Britton et al. 2014; Flook et al. 2010; Jensen and Kenny 2004; Kratter 1983; Moretti-Altuna 1987; Schonert-Reichl et al. 2015; Semple et al. 2010; Telles et al. 2013; Verma et al. 1982, Verma et al. 1982) based on our effect size analysis.

Table 4 Included studies’ effect sizes for attention and executive function
Full size table

The five studies that found an intervention effect provided data for a total of 28 outcome variables of attention or EF, of which 11 showed significant intervention effects. To determine if these significant intervention effects were clinically significant, the mean differences between groups were inspected and if the mean differences between groups were larger than the pooled standard deviation for that measure, then it was considered as clinically significant (Kendall and Sheldrick 2000).

Seven of these 11 significant outcome variables were from computerized measures, namely, the Attention Network Test (ANT; Felver et al. 2014) and the Test of Variable of Attention (TOVA; Sidhu 2013). The significant ANT variables were specifically conflict monitoring (Felver et al. 2014) as well as the overall accuracy and intra-individual coefficient of variation (Leonard et al. 2013). Three out of the 11 variables were from objective measures such as pen-pencil tests which consisted of the Trail Making Test A and B (Purohit and Pradhan 2016) and the Dortmund Attention Test (DAT; Haffner et al. 2006). The remaining one other significant outcome variable was the attention deficit symptoms subscale from the subjective measure FBB-HKS (Haffner et al. 2006).

Two of the 13 included studies had data that was able to be pooled into the a meta-analysis investigating effects of Yoga on EF outcome as measured by the Stroop test (Purohit and Pradhan 2016; Telles et al. 2013) (see Fig. 4). The meta-analysis did not reveal a significant treatment effect for the Word condition (95% CI −0.36 to 0.27; p = 0.78), the Color condition (95% CI −0.44 to 0.19; p = 0.43), or the Color-Word condition (95% CI −0.48 to 0.15; p = 0.30) from the Stroop test.

Fig. 4
figure 4

Meta-analysis of EF as measured by Stroop test post-mindfulness-based intervention for children 8–13 years old (Telles et al. 2013) and orphans 11–16 years old (Purohit and Pradhan 2016)

Full size image

Out of the 13 studies included in this review, only two studies reported using a Mindfulness outcome measure, namely, the Cognitive and Affective Mindfulness Scale Revised (CAMS-R; Britton et al. 2014) and the Mindful Attention Awareness Scale for Children (MAAS-C; Schonert-Reichl et al. 2015). Only one of these studies found a significant intervention effect for Mindfulness, but it also found no intervention effect for attention or EF (Schonert-Reichl et al. 2015). Mindfulness outcomes of the included studies were tabulated in Table 5.

Table 5 Included studies’ effect sizes for Mindfulness
Full size table

Exploring Study Characteristics

The characteristics of the included studies were inspected visually to see if there were any observable patterns or relationships between study characteristics and the results found from the analysis and effect size calculations performed in this review. Visual inspections did not find any obvious patterns or relationships between study characteristics and findings. Further statistical analyses were conducted to explore this in more detail. A chi-square analysis revealed that there were no relationships between the type of interventions, namely, Mindfulness-Based Psychological Intervention, Yoga and Traditional Meditation, and significant findings χ 2 (2, N = 13) = 2.94, p = .23. The relationships between the types of variable, namely, attention vs EF outcomes, was marginally significant, χ 2 (1, N = 21) = 3.23, p = .072; the relationship was trending towards attention outcomes and significant findings.

Other study characteristics explored included the total dosage time of the intervention, the total number of participants, and the study’s methodology quality score on the PEDro. In studies where the dosage time was a range, the average time within the range was used for analysis. The scores on the PEDro were treated as a continuous variable, and the distribution was normally distributed. The Mann-Whitney test indicated that there were no significant differences between the studies with significant findings and the studies with non-significant findings regardless of the dosage of the intervention (U = 13.00, p = .465), the number of participants in the study (U = 17.00, p = .935), or the methodological quality as reflected in the PEDro scores (t(11) = −0.03, p = 0.98).

Discussion

Five of the 13 reviewed studies of mindfulness-based interventions demonstrated efficacy in improving aspects of child and adolescent attention or EF outcomes, with the efficacy independently confirmed by calculated effect sizes (range 0.30–32.03). Overall, the efficacy of mindfulness-based interventions for enhancing attention or EF in children and adolescents remains to be established. The results to date are promising, especially coupled with a systematic literature review showing efficacy of mindfulness practices at enhancing cognitive abilities in adults (Chiesa et al. 2011). Further high-quality research in children and adolescents is needed.

The five studies that found significant effects for attention and EF were based on different facets of attention and EF (Felver et al. 2014; Haffner et al. 2006; Leonard et al. 2013; Purohit and Pradhan 2016; Sidhu 2013). For example, the included studies found significant effects for sustained attention (Sidhu 2013), conflict monitoring (Felver et al. 2014), inhibition and switching executive abilities (Purohit and Pradhan 2016), and overall attention performance (Haffner et al. 2006; Leonard et al. 2013). Most of these significant effects were predominantly based on attention tasks that require visual attention. It is unclear, however, whether mindfulness-based interventions improve specific aspects of attention or EF.

The five studies that found a significant intervention effect for attention or EF did so with mostly quantitative computerized outcome measures (Felver et al. 2014; Leonard et al. 2013; Sidhu 2013). This suggests that the type of assessment used may be crucial to detecting attentional effects of mindfulness. One possible explanation for this may be that computerized assessments are more sensitive to change. Most computerized tests capture the speed of responses (reaction time) measured in milliseconds, making it possible to detect very mild changes in an average reaction time (Collie et al. 2001). In contrast, many pen-paper neuropsychological tests of attention are measured based on accuracy, which means that the maximum level of performance would depend on the number of responses required for that particular test (Collie et al. 2003). As a consequence, most pen-paper tests would have fewer possible levels of performance compared to the possible levels of performance from a computerized test measuring reaction time. An example of this difference in the sensitivity to detecting change between computerized and pen-paper tests in a pediatric population was illustrated in a study that examined the attentional processes in children treated for cancer (Butler and Copeland 2002). This study demonstrated that the computerized attention measure, namely, Conner’s Continuous Performance Test, was the most sensitive to measuring change in attention (estimated effect size d = 0.84) compared to the other pen-paper attention measures they selected which only showed moderate levels of change (Digit Span, d = 0.48; sentence memory, d = 0.55; Butler and Copeland 2002; Raskin 2011). On the basis of this review, future studies should consider including a computerized assessment task for enhanced sensitivity. Further, psychophysiological measures of attention such as those explored in adult literature, for example, measures of attention using EEG and ERP (Moore et al. 2012), using the attentional blink paradigm (Slagter et al. 2007; Slagter et al. 2009), or measuring eye movements (Oken et al. 2006), should be considered for the pediatric population as well. Psychophysiological measures have the potential to measure ones’ continuous attentional state (Vanhala et al. 2006) and, in recent studies, the ability to obtain functional brain networks of attentional performance (Rosenberg et al. 2015). For example, physiological changes would be more sensitive to detecting the exact reaction response (e.g., reaction time, pattern, and behavior) to an unexpected stimulus (Vanhala et al. 2006).

Another characteristic of the studies that could not be explored in this review is who delivered the mindfulness-based interventions. This characteristic could not be explored because not all the studies reported this information. This characteristic may be crucial in determining what outcomes the intervention may find. For example, if the yoga intervention was developed and delivered by a psychologist, then the intervention may have more of an emphasis on psychological and emotional well-being and mindfulness. Conversely, if the yoga intervention was delivered by a physiotherapist, then the intervention may have more of an emphasis on physiological aspects such as more focus on posture and alignment; and if the intervention was delivered by a teacher, the intervention may have more of an emphasis on behavior. These types of emphasis, mostly due to disciplinary bias, are not well reported in protocols of studies. The findings of studies may depend on what content was emphasized, participants’ desires, and the intervention content being targeted. Future studies need to clarify who developed and delivered the intervention and if there was emphasis on particular aspects such as attention, psychological well-being, physical improvements, or behavior.

The second aim of this study was to examine mindfulness outcomes within the included studies. Unfortunately, only two of the 13 included studies reported Mindfulness as an outcome measure, despite the fact that all of the 13 included studies were trialing a mindfulness-based intervention. The two studies that reported a Mindfulness outcome, including the study that reported a significant effect for Mindfulness, did not find a significant intervention effect for attention or EF. The paucity of mindfulness assessment within this literature makes the interpretation of the results challenging as it remains unclear whether or not the interventions tested sufficiently improved mindfulness as it is usually measured in research and clinical practice.

Mindfulness-based interventions may need to specifically target attention to have an effect on attention. In the 13 included studies, four of them specifically targeted attention as their primary outcome (Felver et al. 2014; Leonard et al. 2013; Semple et al. 2010; Sidhu 2013), four of the studies targeted EF (Flook et al. 2010; Purohit and Pradhan 2016; Schonert-Reichl et al. 2015; Verma et al. 1982), while three of the studies primarily focused on all ADHD symptoms which consisted of attention and executive outcomes. The primary focus of the two remaining studies were physical fitness (Telles et al. 2013) and acceptability of mindfulness (Britton et al. 2014). In addition, heterogeneity in the types of mindfulness-based interventions used may account for lack of clarity within the current literature. This suggests that it may be necessary to carefully test standardized mindfulness-based intervention protocols, such as Mindfulness-based Stress Reduction (MBSR) (Kabat-Zinn 1991). MBSR has been used in multiple studies and research on a range of different conditions and populations (Cramer et al. 2012; Gotink et al. 2015; Hughes et al. 2013; Khoury et al. 2015; Lao et al. 2016; Ledesma and Kumano 2009; Niazi and Niazi 2011; Parswani et al. 2013; Praissman 2008; Rosenzweig et al. 2010).

Mindfulness-based training in the existing literature all consist of body awareness training (e.g., observing the breath) where people focus internally on their own bodily sensations and state. Yet, the majority of the attention outcomes reviewed in this paper consisted of (1) monitoring how well participants attend to an external stimuli—that being a stimuli on paper (n = 8) or on a computer screen (n = 5); and/or (2) attention ability in real life as reported by parents or teachers (n = 7). Improvements in attending to internal bodily states would be difficult to observe by another person so these measures too are measures of attention to external stimuli. This is potentially problematic. If improving attention to external stimuli is the goal, then the object of focus for mindfulness should be external stimuli. Alternatively, if improving attention to internal stimuli is the goal then that should be measured, such as measuring participants’ physiological responses (e.g., EEG) while they perform a mindfulness practice observing their breath.

In addition to limitations already discussed, many of the studies had small sample sizes, limiting power to find significant effects on attention. Further, the existing research is heterogeneous in terms of participation, participant populations, assessment measures, and types of intervention, making it challenging to draw clear conclusions.

Future studies in this area may enhance the evidence base for mindfulness-based intervention in children and adolescents with more rigorous experimental design. Randomized controlled trials with larger sample sizes, mindfulness trainings that are compatible with the focus of the outcome measures (e.g., mindfulness trainings with a focus on external stimuli if the outcome measure assesses attention to external stimuli), and having a standardized intervention are recommended additions to the experimental design. Further, the additions of blinded and sensitive outcome measures, such as computerized or psychophysiological measures of attention, to future trials would help avoid biases and strengthen the results. Although a conclusive evaluation cannot be drawn from the existing literature, this review suggests that mindfulness-based interventions are promising. High-quality studies are required in pediatric populations with attention or EF problems other than ADHD, such as children with cerebral palsy (CP) or acquired brain injury (ABI), as the findings will help determine the use of mindfulness-based interventions in clinical and rehabilitation settings.

The effects of mindfulness-based interventions on attention and EF in children and adolescents cannot be clearly concluded from the current literature; however, there is promising data, indicating the need for future research. Further high-quality studies focusing on standardized mindfulness-based interventions and using standardized attention measures are needed.