Parenting Stress in Families of Children With ADHD

Search and Retrieval of Studies and Eligibility Criteria

A computerized search was conducted to locate relevant studies using PSYCInfo, ERIC, Medline, Dissertation Abstracts International, and Google Scholar. Search terms were composed of combining descriptors addressing ADHD (attention deficit disorder with hyperactivity, attention deficit disorder [ADD], hyperactivity, hyperkinesis, minimal brain dysfunction, behavior problems) with descriptors addressing parenting stress (stress and caregiver burden; see Note 2). The citation indices of PSYCInfo, ERIC, and Medline were also searched for reports citing those reports already identified. The reference lists of relevant identified articles were reviewed for appropriate studies not identified elsewhere. In addition, review articles (e.g., Fischer, 1990; Johnston & Mash, 2001), books (Deater-Deckard, 2004), and a listing of relevant articles produced by the developer of the dominant parenting stress measurement tool (Abidin, 2003) were consulted. Using the title and abstracts of the reports, this body of work was narrowed down to 117 studies for possible inclusion in the meta-analysis.

Studies were included in the meta-analysis based on the following criteria:

Study Characteristics and Coding

A total of 44 written reports (22 journal articles and 22 dissertations) met the criteria for inclusion. In some cases, two or more written reports were available on the same data. Articles from peer-reviewed journals were preferred; dissertations and theses were used if no formally published reports were available. This hierarchy was established to ensure that the highest quality reports were included in the meta-analysis and to ensure that the most comprehensive data were available. These 44 studies generated 208 effect sizes on a total sample of 4,991 families.

A coding manual was developed for the coding of study reports. This coding scheme was based on the research questions formulated above and on an examination of common data points present within this body of literature. Primary coding was completed by the author using pencil-and-paper forms and entered into Comprehensive Meta-Analysis Version 2.2 (CMA; Borenstein, Hedges, Higgins, & Rothstein, 2007) for analysis. The coding scheme required little interpretation. Study descriptors coded were bibliographic information (citation, publication year, publication type), methodological information (ADHD definition, ADHD diagnostic procedures, measure used), and sample information (percentage female children in sample, percentage mothers in sample, mean child age in sample, mean SES of sample). Effects were coded for their relationship with each of total, parent domain, and child domain stress on the PSI (and the PSI–Short Form [PSI/SF], where parental distress was equated with parent domain stress from the PSI, and difficult child was equated with child domain stress from the PSI), the most common measure of parenting stress. Results from measures other than the PSI were coded as total stress, and measures that included evaluations of both intensity and frequency were coded for their measure of intensity if both were available.

To ensure independence between comparable effects, only one measure of parenting stress was used from each study (e.g., PSI or Family Strain Index, not both; however, total, parent, and child domain stress scores from the PSI were all used where applicable as each was entered into a separate analysis). This decision was made systematically using the order of preference laid out in the coding manual (with results from the PSI being the most preferred—in all cases where more than one measure was available, this was the one used). This hierarchy was developed based on the prevalence of the measures in the literature, such that more prevalent measures were always given preference so that niche measures with little empirical support would not dominate the analysis. When results were present from more than one parent on each child, the results from the mothers were used, as father data were often obtained on only a subsample of participating children in any given study. In only one journal article (Peris & Hinshaw, 2003) were insufficient data available for statistical analysis. The primary author of this study was contacted but did not provide the necessary data. (This study was not included in the meta-analysis and was not counted in the 44 studies listed above.)

Analysis

The methodology recommended by Lipsey and Wilson (2001) was used for this meta-analysis. Separate meta-analyses were conducted to answer each of the research questions as described above. Cohen’s d, representing the standardized mean difference, was used for the analyses that involved group comparisons (except for the mother–father comparison, where the standardized paired difference was used). The pooled correlation coefficient, r, was used for dimensional (correlational) analyses. Effect sizes were interpreted as suggested by Cohen (1977, 1988): for d: ≤.20, small; .50, medium; and ≥.80, large and for r: ≤.10, small; .25, medium; and ≥.40, large. Given the large number of analyses conducted, an alpha level of .01 was used in inspecting the weighted effect sizes and moderator analyses to reduce the probability of Type I error. Given the number of analyses performed, this likely still resulted in an elevated rate of Type I error; however, this alpha level reduced the studywide probability of making a Type I error from 86% to 32%, while maintaining the risk of a Type II error at a reasonable level. Using the Bonferroni correction with a studywide alpha level of .05 would have required each analysis to have alpha set at .001. Specific p values are provided in text and tables to permit consideration of the findings from a stricter or more lenient alpha level for interested readers. Mean effect sizes were considered interpretable if composed of 3 or more effect sizes.

For each analysis, a Forrest plot was created using CMA and inspected visually for outliers (Borenstein et al., 2007; Johnson & Eagly, 2000; Lipsey & Wilson, 2001; Sterne & Harbord, 2004). An examination of all outliers was also undertaken using the often employed criterion of individual study effect sizes exceeding three standard deviations from the composite effect size (Lipsey & Wilson, 2001). Using these criteria, outliers were common in this meta-analysis, although in few cases could obvious reasons for the variation be found. Given the relatively small number of effect sizes in each analysis and the ubiquity of outliers in this study, each was retained at its current value. In the majority of cases, however, the outlier effects were in the same direction as the mean effect (and other component effects), suggesting little question about the presence of a given effect.

Homogeneity analyses were conducted on each analysis using the Q statistic (Lipsey & Wilson, 2001). Given that the majority of the analyses reflected heterogeneous effect sizes, a random effects model was consistently used in conducting the meta-analysis (Lipsey & Wilson, 2001). In a few analyses (e.g., the correlation between parent domain stress and parental depressive symptoms), the null hypothesis of homogeneity was not rejected. As analyses tended to involve a very small number of effect sizes (3), Q may not have had sufficient power to detect an effect of heterogeneity, leading to a high chance of Type II error. Therefore, heterogeneity was assumed for all analyses and a random effects model was employed. Furthermore, a random effects model was the logical choice for analysis given the criteria laid out by Borenstein and colleagues (2007) of using a random effects model for studies where there is little reason to believe that the component studies are virtually identical. A random effects model, as opposed to a fixed effects model, assumes variation among effect sizes above sampling level error alone, suggesting the presence of moderators. In a random effects model, each study included in a meta-analysis is assumed to be a random sample of the distribution of effects. Thus, the combined effect is the mean of the distribution of effects. In a fixed effects model, conversely, there is assumed to be one true effect that is shared by all the studies included in the meta-analysis. These models weight individual studies differently. In a fixed effects model, each study is weighted by its sampling error, such that larger studies are weighted more heavily. In a random effects model, each study is assumed to be estimating a different effect size and thus the weights given to each study are more balanced and less influenced by sample size (sampling error).

Separate metaregressions were used to examine the effects of the following continuous moderator variables: child gender (percentage of child sample female), child age, and publication year. Analog ANOVAs were used to evaluate the categorical moderators (ADHD diagnostic criteria, publication type, measure of parenting stress). Analog ANOVAs group effect sizes into mutually exclusive categories based on the variable of interest and then test whether the between-category variance is significant (Lipsey & Wilson, 2001). Mixed effects models were used for both the metaregressions (method of moments) and analog ANOVAs. Due to the small K (number of effects) in each meta-analysis, results were not broken down by significant moderators for each analysis, nor were moderator analyses conducted on each analysis; instead, the correlation between ADHD symptoms overall and total stress was evaluated as a representative illustration of the effect of various moderators on the relationship between child ADHD and parenting stress.

Reliability coding was completed by a master’s level graduate student trained by the author in meta-analytic procedure. Training consisted of four sessions where the principles of meta-analysis, the purpose of the present analysis, and the nature of the coding scheme were discussed. The second coder coded sample studies (not included in the reliability analysis) with discussion and feedback until acceptable agreement was reached on all constructs. The second coder then independently coded a randomly selected sample of 12 studies used in the meta-analysis. Interrater reliability was calculated for each variable in the coding system using Cohen’s kappa for categorical variables and the intraclass correlation coefficient for continuous variables. The agreement for all constructs was high with kappas ranging from .8 (measure used in correlational analysis) to 1 (publication type, ADHD criteria used, parenting stress outcome measure, group matching) and intraclass correlation coefficients ranging from .91 (percentage children female) to 1 (all effect size variables).

Study Characteristics

Forty-four studies met criteria for inclusion in this meta-analysis. The year of publication ranged from 1983 to 2007. Thirty-eight of the studies were conducted in North America, 4 in Western Europe, and 2 in Asia. Most studies used the Parenting Stress Index (Abidin, 1983, 1990, 1995) or a variant of it to measure parenting stress (PSI: n = 20; Parenting Stress Index–Short Form: n = 12; translations of the full or short form: n = 3). The remaining studies used adaptations of the PSI (n = 2; i.e., creating a new stress variable based on a principal components analysis of the PSI and other stress or parenting competence measures), the Family Strain Index (n = 1; Riley et al., 2006), the Circle of Stress Index (n = 1; Circle, 1998), the Disruptive Behavior Stress Inventory (n = 2; Johnson & Reader, 2002), the Parenting Daily Hassles Scale (n = 1; Crnic & Greenberg, 1990), the Nijmegen Child-Rearing Situation Questionnaire (n = 1; Wels & Robbroeckx, 1996, cited in Kadesjö et al., 2002), the Pearlin Parental Stress Scale (n = 1; Pearlin & Schooler, 1978), and the Questionnaire on Resources and Stress–Short Form Adapted (n = 2; Salisbury, 1986).

The studies used different diagnostic criteria to confirm or establish ADHD in the child participants. Nineteen studies used the Diagnostic and Statistical Manual of Mental Disorders–Fourth Edition (DSM-IV) criteria for ADHD, 10 used the DSM-III-R criteria, and 1 used the DSM-III criteria. One study used a combination of the International Statistical Classification of Diseases and Related Health Problems–10th Revision (ICD-10) criteria and DSM-IV criteria, and 1 study of preschool-age children specifically opted not to use organized criteria as the authors felt they were not applicable to that age group (Goldstein, Harvey, & Friedman-Weieneth, 2007). The remaining 12 studies did not specify the diagnostic criteria they used, although all studies included in the meta-analysis employed some measure of ADHD symptoms or a formal diagnosis of ADHD as a condition of eligibility.

The majority of the studies examined parenting stress solely in mothers (n = 24), whereas 9 studies did not report on the parent sample makeup. The remainder of the parent samples were composed primarily of mothers (85% or more of the sample), unless mother–father comparisons were being undertaken. The age range of children in the samples varied from 3 to 12 years, with an unweighted mean age from those studies reporting applicable data of 8.48 years. The percentage of girls in the samples in each study ranged from 0 to 100%, with a mean of 19.25%.

The number of effects comprising each weighted mean effect (i.e., answering each research question) ranged from 3 to 10. Each study provided between 1 and 18 effect sizes (M = 4.73). Results will only be discussed when K ≥ 3. The results of the group comparison studies are presented in Table 1, and results of the dimensional studies in Table 2, in the order of the research questions listed above.

OPEN IN VIEWER

Table 1.

Meta-Analyses of Studies Comparing Groups on Parenting Stress

					95% CI
Comparison	Stress Domain	K	d	p	Lower Limit	Upper Limit	Q
ADHD vs. nonclinical control	total	9	1.80	<.001	1.24	2.37	88.04*
	parent	10	0.90	<.001	0.50	1.30	44.03*
	child	9	2.12	<.001	1.45	2.79	68.62*
ADHD vs. other clinical	total	3	0.30	.385	−0.38	0.98	8.48*
	parent	6	0.17	.313	−0.16	0.51	10.59
	child	6	0.56	.009	0.14	0.97	15.34*
ADHD vs. ADHD + conduct problems	total	5	−0.66	<.001	−0.89	–0.44	4.94
	parent	5	−0.42	.001	−0.67	–0.18	5.73
	child	5	−0.83	<.001	−1.21	–0.45	12.80*
Fathers vs. mothers	total	7	−0.33	<.001	−0.46	–0.19	6.45
	parent	7	−0.06	.608	−0.27	0.16	5.88
	child	5	−0.35	.004	−0.58	–0.11	0.74

Note: Data are coded so that a positive effect indicates that the first group listed reports higher parenting stress than the second group. K = number of effect sizes; d = standardized mean difference, except for fathers vs. mothers comparison, which used the standardized paired difference; CI = confidence interval; Q = test of homogeneity; ADHD + conduct problems = ADHD plus co-occurring oppositional defiant disorder (ODD) and/or clinical levels of aggression.

*

p < .05 (for Q).

OPEN IN VIEWER

Table 2.

Meta-Analyses of Dimensional Parenting Stress Studies

					95% CI
Variable Correlated With Stress	Stress Domain	K	r	p	Lower Limit	Upper Limit	Q
Overall ADHD symptoms	total	7	.51	<.001	.42	.58	14.64*
	parent	4	.28	.005	.09	.45	6.29
	child	4	.50	.047	.01	.79	45.14*
Inattention	total	8	.37	<.001	.25	.49	24.22*
	parent	4	.28	.001	.11	.43	4.45
	child	3	.43	.081	−.06	.76	18.38*
Hyperactivity/impulsivity	total	9	.49	<.001	.41	.56	17.23*
	parent	4	.38	.001	.16	.56	8.48*
	child	3	.63	<.001	.35	.80	9.90*
Conduct problems	total	10	.56	<.001	.41	.68	49.88*
	parent	5	.39	<.001	.27	.50	3.18
	child	4	.66	.005	.24	.88	24.29*
Internalizing	total	4	.57	<.001	.29	.76	10.23*
	parent	3	.47	<.001	.32	.60	1.38
	child	3	.63	.001	.30	.82	5.39
Parental depressive symptoms	total	5	.48	<.001	.26	.66	21.04*
	parent	insufficient data
	child	insufficient data
Marital quality	total	4	−.20	.075	−.40	.02	6.62
	parent	insufficient data
	child	insufficient data

Note: K = number of effect sizes; r = pooled correlation coefficient; CI = confidence interval; Q = test of homogeneity.

*

p < .05 (for Q).

ADHD Diagnosis (Level of Symptoms)

As shown in Table 1, parents of children with ADHD experienced significantly more parenting stress than parents of nonclinical controls in terms of total, parent domain, and child domain stress. These effects were large. As shown in Table 2, the relationship between overall ADHD symptoms and total parenting stress was large and the relationship between overall ADHD symptoms and parent domain stress was medium. The effect was not significant for child domain stress. Breaking this down further, the effect size for the relationship between inattention and total parenting stress as well as parent domain stress was medium, whereas the association between inattention and child domain stress was not significant (see Table 2). Finally, the effect between hyperactivity/impulsivity and parenting stress was large for total stress and child domain stress but medium for parent domain stress (Table 2).

Parents of children with ADHD were also compared to parents of children with learning disabilities (k = 1), autism (k = 1), developmental delays (k = 1), internalizing disorders (k = 1), and unspecified clinical disorders (k = 2; e.g., consecutive non-ADHD referrals to a children’s mental health agency). These “other disorders” were combined into one analysis. As shown in Table 1, these comparisons were only significant in the area of child domain stress, where the effect was medium.

Co-Occurring Disorders

As shown in Table 1, parents of children with ADHD and clinical levels of co-occurring conduct problems (defined as ratings of conduct problems, oppositionality, externalizing behavior problems, and aggression and/or clinical diagnoses of conduct disorder or oppositional defiant disorder) experienced significantly more parenting stress than parents of children with ADHD who did not have conduct problems. The effect size was medium for total stress, small for parent domain stress, and large for child domain stress. As shown in Table 2, the size of the relationship between total and child domain stress and conduct problems was large, whereas the size of the relationship between parent domain stress and conduct problems was medium. The association between internalizing symptoms and parenting stress in children with ADHD was large for total, parent domain, and child domain stress.

Parent and Contextual Factors

As shown in Table 2, a large effect was found for the association between parental depressive symptoms and total stress (insufficient data were available for interpretation on the relationship between parental depressive symptoms and child and parent domain stress). The effect size for fathers versus mothers for total and child domain stress was small but negative, indicating that mothers experience more stress than fathers, as shown in Table 1. However, the results did not achieve significance in the parent domain, indicating that mothers and fathers of children with ADHD do not differ significantly in their reports of parenting stress in this area.

The variable “marital quality” was composed of measures of both marital satisfaction and marital discord, which was reverse-scored. The marital discord effect, once reversed, was similar in size to the marital satisfaction effects, supporting its inclusion here. The association between marital quality and parenting stress was not significant.

Moderator Analyses

As stated earlier, all moderator analyses were conducted on the effect for the association between ADHD symptoms and total stress (Fisher’s Z, a transformation of r, was used in these analyses rather than r, as the sampling distribution of r is not normally distributed). Child gender (percentage child sample female) was a significant moderator, Q_R(1) = 9.77, p = .002, with less parenting stress in samples with higher proportions of girls. Mean child age, however, was not a significant moderator, Q_R(1) = 0.54, p = .463.

There were no significant methodological moderators. Although the effects for publication type (journal vs. thesis/dissertation) showed a trend toward larger effects in the published works (r = .55) than the unpublished works (r = .42), the difference was nonsignificant, Q_B(1) = 2.21, p = .14. A weighted regression analysis showed that publication year was not a significant moderator, Q_R(1) = 2.59, p = .11. The ANOVA analog for the parenting stress measure was not significant, Q_B(4) = 2.43, p = .66. Finally, diagnostic criteria used in designating the sample (e.g., DSM-III, DSM-IV, other) were not a significant moderator, Q_B(2) = 2.52, p = .28.

Strengths and Limitations

One element of this meta-analysis that could be considered both a strength and a limitation is the inclusion of unpublished studies. The strength of this approach is that it avoided the “file drawer” problem common to studies with null effects. On the other hand, unpublished studies cause greater concern in terms of methodological quality as they have not passed peer review. The eligibility criteria set out for this meta-analysis, however, excluded those studies with truly questionable methodology or analysis. Given the constraints of current meta-analytic technique, however, the relative statistical simplicity of these studies was not detrimental as the results of t tests and correlations are easily meta-analyzed, whereas results from sophisticated published studies, which included structural equation models and regressions, are not (Lipsey & Wilson, 2001). Indeed, the inability to include all the results from some of the more sophisticated studies on parenting stress and ADHD (e.g., Bussing et al., 2003; Harrison & Sofronoff, 2002; Peris & Hinshaw, 2003) was a limitation of this meta-analysis.

Another limitation to this meta-analysis was the lack of a searchable keyword or “descriptor” for parenting stress in any of the relevant article databases. The absence of parenting stress as a descriptor is surprising given the sizable research on this topic; however, without it, one can never be entirely confident that all relevant studies were uncovered during the retrieval phase.

The limited number of component effect sizes (k) in some analyses is also a limitation of this study. A number of the analyses included only 3 to 4 studies. Although meta-analysis is still statistically appropriate in these cases, one study may sway results. A random effects model was used, however, to mitigate this possibility. The results of analyses with a limited number of effects may be fruitful areas for further inquiry in order to confirm the findings.

The preponderance of mothers in the parent samples included in the meta-analysis is another limitation. The findings from this meta-analysis are therefore more readily applicable to mothers than fathers, although given the generally small differences between mothers and fathers found in this study, it is likely that the results apply to fathers as well. One final limitation was the lack of consideration of sample characteristics, including referral sources (parent training programs, clinical vs. community samples).

Directions for Further Research

Although the literature on parenting stress and ADHD is reasonably well developed, this meta-analysis highlighted several areas for further inquiry. The considerable research base in this area and the magnitude of this effect indicate that further research on parenting stress and ADHD should focus on the mechanisms of this relationship, rather than on establishing again that these parents are more stressed than parents of typically developing children. A major limitation of the literature to date is that the child participants in all of the studies were 12 years of age or younger, possibly because the most commonly used measure of parenting stress (the PSI) is normed for this population. Although the manifestations of ADHD in adolescents differ somewhat from those in younger children (i.e., reductions in hyperactivity), adolescents are apt to engage in behaviors such as risky driving and substance use (Barkley, Fischer, Smallish, & Fletcher, 2004) that may indeed be associated with elevated parenting stress. Research on parenting stress in parents of adolescents with ADHD is therefore clearly warranted.

Second, future research needs to further explore parental and contextual predictors of parenting stress in families of children with ADHD. The few parental and contextual variables in this meta-analysis with sufficient data showed significant effects. One variable in particular that warrants further attention is parental ADHD, especially given the relationship between adult ADHD and depression (e.g., Biederman et al., 1992) and the findings here of a positive relationship between parental depressive symptomatology and parenting stress. Parental cognitive factors, such as parenting sense of competence, have also received very little attention in spite of Mash and Johnston’s (1990) model giving maternal cognitions a substantial role in parent–child interactive stress. Parental education and age are other variables requiring further exploration. Similarly, non-nuclear family arrangements require consideration, along with consideration of family size. Cultural and language differences also bear exploration. Buffers such as social support and marital status also need further investigation, as these may be important factors to consider in interventions for elevated parenting stress.

Although the literature on child-level factors and parenting stress in families of children with ADHD is generally larger than that on parent and contextual-level factors, the effects of ADHD subtype require more attention, as do the effects of co-occurring internalizing and learning disorders.

Implications and Conclusions

This study clearly illustrated that it is important for clinicians to consider parenting stress in families of children with ADHD. Although moderate stress is adaptive, high stress has negative implications for both physical and mental health. Parents experiencing high levels of parenting stress are less likely to be capable of implementing interventions that change children’s behavior (Kazdin, 1995). As a consequence, clinicians, school personnel, and policy makers need to be aware of the high parenting stress of parents of children with ADHD and take necessary steps to reduce this stress.

Within the school setting, this research has substantial implications. Parents of children with ADHD are frequently asked to support their children’s learning (Rogers, Wiener, Marton, & Tannock, 2009a) because of the academic underachievement associated with this disorder (Loe & Feldman, 2007). The findings from the current research indicate that due to elevated rates of parenting stress, these parents may not be able to adequately support their children in this way. In addition, Rogers, Wiener, Marton, and Tannock (2009b) have shown that elevated levels of parenting stress are associated with less supportive styles of parental academic involvement. These requests may therefore serve only to increase parenting stress further, without positive effects on a child’s learning, especially in families of children with both ADHD and co-occurring internalizing or externalizing difficulties.

From the perspective of policy, this research has implications for community-wide resources and funding. The severity of parenting stress in families of children with ADHD is so substantial that it suggests that these parents require greater support in coping and managing than is currently provided. Many jurisdictions provide free or reduced-cost respite care to parents of children with autism and other developmental disabilities. Although their stress is considerable, parents of children with autism do not experience higher levels of parenting stress than parents of children with ADHD (Markham, 2001). It is likely that parents of children with ADHD would benefit from similarly funded services. Other considerations such as programs to address parenting stress within the confines of children’s mental health programming are needed. Currently, this type of service is typically considered to fall under the realm of adult mental health services, which many families may not be accessing. To the extent that parenting stress affects parenting practices, as suggested by Abidin (1990), McCleary (2002), and Webster-Stratton (1990), intervening in parenting stress has the potential to achieve far-reaching benefits within families and communities and thus is an appropriate target for agencies targeting child and family well-being. Expanding on this ideal service delivery model further, the effect of parental depressive symptomatology on parenting stress suggests that integration of mental health services across the lifespan would be helpful for families. The current dichotomization of services by age requires families and clinicians to either ignore some factors contributing to family struggles, attribute those struggles to an inaccurate source (the identified client: the child with ADHD), or try to integrate services from multiple agencies.

Clinically, therapists should be attuned to the needs of parents when addressing child ADHD. A comprehensive family approach, such as the interventions for parenting stress designed by Kazdin and Whitley (2003) and Treacy, Tripp, and Baird (2005), is suggested by this research. Both of these interventions for parenting stress focus on the role of the parent, including instruction in problem-solving skills, self-care, and support in managing a variety of stressors in the parent’s life. Furthermore, parent training programs, such as COPE (Cunningham, 2006), may wish to address parenting stress more directly, although this program does already have positive effects on parenting stress (Thorell, 2009). In addition, clinicians should assess the multiple potential risk factors in these families for increased parenting stress (i.e., parental depression, child internalizing and externalizing symptomatology). There are few differences between mothers and fathers in parenting stress; thus, both parents should be targeted for interventions for parenting stress. The pervasiveness of the PSI, which was designed as a clinical tool (Abidin, 1995), in conducting research into this topic also showed how simply and practically this construct can be assessed by clinicians.

In summary, this meta-analysis, the first quantitative review of parenting stress and ADHD, confirmed findings of higher parenting stress in parents of children with ADHD than parents of children without ADHD. Severity of ADHD symptoms was linked to severity of parenting stress and co-occurring conduct problems and internalizing symptoms significantly increased parenting stress. Parent and contextual factors also played a role in parenting stress, with parental depressive symptomatology showing the largest effect. Child gender was a significant moderator of the relationship between total parenting stress and ADHD symptoms. By systematically collating the data on this topic, this meta-analysis was able to identify areas requiring further research and suggest new areas of inquiry. The magnitude of these findings underlines the importance of considering parenting stress in research, clinical work, schools, and policy addressing the needs of families of children with ADHD. Future research should be directed at exploring parent and contextual factors further, considering the parenting stress of parents of adolescents, and evaluating interventions directed at reducing parenting stress in parents of children with ADHD. Given the numerous negative outcomes associated with elevated parenting stress (reduced response to intervention, decreased parental psychological well-being, constrained child development), this will be critically important for the well-being of families of children with ADHD.