Childhood ADHD and treatment outcome: the role of maternal functioning

Maternal ADHD, attachment style, and resilience

The etiology of ADHD is multifactorial, as both genetic and environmental factors have been evidenced in the development of ADHD [5, 6]. For example, a relative with ADHD [7], an increase in Copy Number Variation (CNV, [8]), prematurity [9] and some form of neglect (Thapar et al. 2012) have all been implicated. Moreover, numerous studies have found ADHD to be associated with a poor prognosis (e.g. more divorces, higher rates of substance abuse disorders in adulthood, and increased mortality rate) [11–13]. Furthermore, the prognosis of ADHD worsens in the presence of comorbidity [11, 14]. Parent and child ADHD are found to be significantly associated, as well as parent and child conduct problems [15]. However, factors associated with the developmental progression and the long-term prognosis of ADHD are not fully understood [15–17] and hence, a greater focus on the developmental progression of ADHD is required. Whereas studies converge in proposing that parental psychopathology poses a high risk of transmission to offspring, relatively little is known about the role of maternal functioning in terms of attention deficit hyperactivity disorder (ADHD) symptoms, attachment style and resilience as predictive factors for treatment outcome when offspring are diagnosed with ADHD.

Maternal ADHD may be a potential risk factor in the development of offspring ADHD [18–21]. Moreover, in a recent study on associations between parental psychiatric disorders and offspring ADHD, maternal diagnosis showed stronger associations with child ADHD than paternal diagnosis [22]. Despite research establishing a link specifically between parental ADHD and parenting, measured as the level of home chaos, and parenting practices assessed through self-reports [23], to date there have been no studies investigating the role of maternal ADHD symptomatology as a prognostic factor in offspring treatment outcome.

In addition to the contribution of maternal ADHD symptoms to the increased risk of offspring developing ADHD, the quality of the mother–child relationship has also come under scrutiny. Indeed, both ADHD and attachment have been proposed as risk factors [24]. Reviewing the literature on ADHD and attachment, we found a clear association between ADHD and insecure attachment. When one condition was present, this increased the risk of developing the other. This underlines that ADHD and insecure attachment may constitute mutual risk factors [25]. In a recent comprehensive review of parental self-reported attachment style and caregiving, adult attachment security was consistently associated with more positive parenting whereas insecurity was related to more negative parenting [26]. These findings underscore the importance of investigating maternal ADHD-symptoms and attachment style as part of a broad assessment of maternal functioning. Alongside the contribution of adult attachment to maternal functioning, a growing body of research suggests that resilience is also a key factor, with greater resilience associated with psychological adaptation and functioning in the face of adversity [27–29]. Broadly, resilience theory focuses on understanding healthy development in the face of risk, and on strengths as opposed to weaknesses. It has been defined as a “pattern of positive adaptation in the context of past or present adversity” [30]. Critically, resilience does not suggest the absence of adversity or risk, but rather highlights the presence of protective processes leading to healthy adaptation. Whilst definitions and measurement of resilience vary considerably from study to study and the scientific value of the concept generally has been debated and challenged [31] resilience has been found to influence treatment response across different manifestations of adversity. These include chronic illness, psychiatric disorders and school bullying [32, 33].

In this study, the focus is on the significance of maternal resilience for parenting in situations where the child needs extra support.

The current study

The aim of the present study was to examine associations between mothers’ functioning and treatment response in their children diagnosed with ADHD who are receiving care as usual.

We hypothesized that maternal self-reported ADHD symptoms; self-reported attachment style and maternal resilience would all be significantly correlated with treatment outcome. More specifically, we anticipated the following; (1) higher maternal ADHD-symptom scores would be associated with lower ADHD symptom reduction in offspring within the first year of treatment. (2) Higher scores for self-reported anxiety or ambivalence on the attachment style questionnaire are associated with lesser symptom reduction in offspring. (3) A higher degree of self-reported maternal resilience is associated with better treatment response in offspring diagnosed with ADHD.

Participants

The families participating were recruited from two child psychiatric outpatient clinics in Region Zealand, Denmark. The four interviewers participating were the same at the two sites; two conducting maternal attachment interviews and two conducting child attachment interviews. Sixty-seven (N = 67) child-mother dyads were included in the follow up.

Maternal measures

Child measures

Procedure

Initial identification was based on referral diagnosis based on ICD-10 (ADHD: combined-type, inattentive-type and hyperactive/impulsive-type) followed by a clinical evaluation in order to establish whether the family met the inclusion criteria.

In order to lower the risk of selection bias, we applied consecutive recruitment, as all the families with children in the appropriate age range presenting the relevant referral diagnosis, were invited to participate.

The data obtained through questionnaires and interviews in the two-stage inclusion process and a 1-year follow up period were analysed using structural equation modeling and mixed effects modeling for repeated measures (Fig. 1).

During the recruitment phase, 203 children were referred with a possible diagnosis of ADHD. Only children in the age range of 7–12 representing middle childhood were subsequently invited to participate in the study. Exclusion criteria further included children who were adopted or living in foster care and children with major handicaps, such as hearing impairment or learning disabilities, preventing them from participating in and completing the interviews. Children who during the inclusion procedure were suspected of psychosis or who had an IQ below 70 were excluded.

Those who met the above criteria were invited to participate and were given further information about the study in a telephone call by the 1st author. The second stage of inclusion involved a formal assessment of ADHD diagnosis, using the Schedule for Affective Disorders and Schizophrenia for School-aged Children, Present and Lifetime Version (K-Sad-PL) [44, 45].

Only patients assigned an ADHD/ADD diagnosis according to DSM-V with ADHD (314.01) or ADD (314.00) in step two were included for follow up. The included families attended both a clinical and a research track separated from each other (Fig. 2).

Apart from exchanging information on treatment initiation, child IQ and in the event of drop out from medical treatment, the research project was separated from the clinical track.

Follow up

The families progressed to the follow up 3 months after the initial assessment (Fig. 3).

This delay was to ensure that the assigned treatment had been initiated. The attachment assessment was conducted at the child psychiatric clinics by four trained interviewers. The interviewers were not involved in treatment and had no prior knowledge of the families. The two interviewers of the mothers were blinded to results of the child assessment and vice versa.

The follow up consisted of four ADHD-symptom screenings completed every third month within the first year of inclusion. The ADHD-RS was sent to the parents for assessment of ADHD-symptom load in their children. The observed change (slope) in ADHD-RS from baseline to the 12 month follow up represents the primary outcome’.

In one-parent households, the schedule was sent to the household in which the child was residing. More than half (53%) of the families were one-parent households and only in one family was the child residing with the father.

Data analysis

The ECR, RSA and ASRS scores used are latent variable predictions derived from structural equation models. ECR and RSA items are modeled assuming multivariate normality. ASRS items are considered ordinal items and modeled using a logit link function. All three latent variable predictions are standardized with means at 0 and variances fixed to 1. The latent variable approach has the advantage of removing the residual (unexplained by the latent factor) variance for each item when predicting the latent variable. Measurement error associated with psychometric scales is minimized in this way.

Because of some missing data on covariates, we used a full information maximum likelihood estimator implemented in Stata 14. Means and variances as well as covariance for exogenous variables were estimated and used as the basis for corrected regression estimates. Missing data due to non-attendance at the follow up measurements were assumed to be random, i.e. not associated with the unobserved outcome measurement. This assumption was substantiated by a missing data analysis showing no significant association between baseline ADHD-RS score and missing data (RR = 1.02, p = 0.47) and between the ADHD-RS score preceding (t₋₁) a missing data incident (RR = 1.01, p = 0.66). Mixed effects Poisson models were used for the missing data analysis.

Predicting ADHD-RS reductions

Methodological considerations

This study is unique in representing the first attempt at exploring the role of maternal functioning in predicting treatment response of children with ADHD; an area which has been lacking in previous studies [15, 16].

The study is, to our knowledge, one of very few studies to address maternal resources as well as risk factors simultaneously and it clearly raises important questions for future research.

Nevertheless, the current study has several limitations.

Our study had a naturalistic design and applied consecutive recruiting in order to include and follow a sample representative of the population we intended to investigate. This was to keep selection bias to a minimum, and hence, we further kept the research track separated from the clinical treatment during the 1-year follow up. However, since a little more than half declined to participate, this then resulted in selection bias. When asked, the participant’s main reason for declining to participate was due to the stress of the many appointments entailed by the clinical track.

Further, another limitation relates to the duration of the follow up period. As factors in the parent–child relationship are likely to emerge over time, a longer follow up would have been ideal.

For example, the lack of correlation between maternal self-reported attachment style and child outcome on ADHD-RS may relate to the relatively short follow up period. This is supported by findings from other studies suggesting that the influence of medical treatment tends to wear off, making other factors more influential in long-term treatment strategies [71, 72].

Another limitation concerns the sole reliance on self-reporting in the assessment of various aspects of maternal functioning as socially desirable responding has been shown to affect results in previous research [73]. However, it has been shown that adults provide an accurate assessment of their own ADHD symptoms [74].

Also, maternal ADHD symptomatology was assessed very specifically; other types of maternal psychopathology were not assessed and thus taken into consideration. For example, as maternal depression is known to be a risk factor for symptom severity in offspring ADHD [19], it would have been of interest to gain information on how many mothers were clinically depressed. We did, however, obtain general information on history of mental illness, from preformed protocols collecting psychosocial and demographic data, which was controlled for in the analysis. The same is also true of the assessment of the children, as the degree of impairment in the children was not assessed beyond an assessment of their ADHD symptoms.

Disappointingly, we were not able to draw any conclusions in relation to the role of fathers’ functioning in the treatment response of their children diagnosed with ADHD. The response rate of fathers was too low to include in the study. Whilst this is not unique to our study (see for example, [75]), previous findings do suggest a link between child externalizing problems and paternal ADHD symptoms, and this may constitute an area of particular interest for future research [54].

Compared to other reported studies, the rate of comorbidity was very low in our study [76, 77]. However, it would appear that ADHD is a clinical predictor of comorbidities, as ADHD seems to increase the risk of developing comorbidity during childhood and adolescence such as conduct disorder and oppositional defiance disorder, which further increase the risk of antisocial behavior and juvenile delinquency later on ([11, 13, 78, 79].This developmental pathway was also observed in a longitudinal study in which children were included only if they had no comorbidity [12]. Yet, in this study by Klein and colleagues, 84 of 135 participants developed probable or definite conduct disorder during adolescence and a further 25% of these developed antisocial personality disorder in adulthood. The children in our study were included after a first time referral and were in the age range 7–12. Hence, the frequency of comorbid disorders may rise in the years to come.