Introduction

Involving parents in the treatment of youth with mental health problems is an important agenda for several reasons. First, parenting practices and behaviors are consistently associated with positive treatment outcomes [5, 7, 11, 22, 28]. Second, post-treatment maintenance of gains is contingent on continued application of the strategies, insights, and skills learnt in therapy [27]. Parents are usually the natural first choice in helping their youth and children maintain gains made in the clinic. Third, a recent study by Hawley and Weisz [10] found that therapists’ alliance with the parents was associated with more family participation and treatment retention in a referred community sample. However, there is little research regarding the extent and nature of parent involvement in the treatment of children and adolescents, particularly from usual clinical care setting.

Attrition research addresses factors associated with termination after having started treatment. A review of attrition studies by Armbruster and Kazdin [4] indicates that attrition is associated with demographic, socio-economic status, family context, parent- and child psychopathology, but these results are not consistent and conclusive. Kendall and Sugarman [19] suggest that the varying degree of involvement and drop-out may be a function of differences between setting and samples. Their results indicated that the level of treatment retention and drop-out for children with internalizing problems was different from what is found in studies on children with externalizing problems. Other studies found that the severity of problem and co morbidity was associated with varying degree of participation in treatment [18, 19]. This indicates that the assessment of the disorder and its severity is an important aspect determining treatment retention.

In therapy with youth there are most often at least two informants—the youth and the parent. Several studies reported low level agreement of clinical information obtained from parents and youth. For instance, Achenbach, McConaughy and Howell [2] found a mean correlation of .25 between parent and child report on self-reported behavioral and emotional syndrome scales. Cantwell, Lewinsohn, Rohde, and Seerley [6] found a range between .19 and .79 for the level of agreement between parent- and youth report of diagnostic information. There was considerable variation between diagnostic categories. There was lower level of agreement for internalizing disorders than for externalizing disorders and disorders with salient somatic symptoms (e.g., core symptoms of anorexia and bulimia).

Discrepancy between parents and youth may be a function of the different context in which parents observe the youth’s behavior compared to the youth’s experience of the problem [2]. Kramer et al. [21] examined responses of 258 referred youths and their parents regarding the type of problems that they agreed on. Generally, they found low-level agreement for areas that involved subjective evaluation of relationships. Interpersonal relationships (or lack of it) could be construed as important for youth—given the centrality of peer relationships for adolescents. Further, the general focus of psychosocial treatments revolves around interpersonal issues [13, 14]. It is, therefore, conceivable that parental (un)awareness of interpersonal problems of the youth would affect their involvement in treatment process.

In a non-treatment context, Ferdinand, van der Ende, and Verhulst [8] followed a cohort of subjects who were 11–14 years at time 1. Direct reports of youth and parents on CBCL/YSR and the discrepancy between their accounts were examined for their ability to predict adverse outcomes, such as, police contacts, suicide attempts, drug use, and referral to mental health services at time 2–4 years later. A series of regression analyses ending in a final forward stepwise regression analysis showed that nine Child Behaviour Checklist (CBCL) syndrome scales, six Youth Self Report (YSR) scales, and six discrepancy scores were significantly related to the outcome measures. In treatment context, discrepancy may be an important source affecting the process of determining treatment goals and the process of realizing these goals.

Although there are clinical studies involving parents, to our knowledge, there is a lack of theoretical and conceptual work regarding parent involvement in clinical work with children and adolescents. There are, however, advances regarding parent involvement in other fields of psychology. For instance, in educational psychology, parent involvement was defined as the dedication of resources given by the parent to the child within a given domain [9]. Grolnick and Slowiaczek proposed a multidimensional construct of parent involvement that could be expressed as first, specific behaviors such as going to school and participating in school related activities. Second, personal involvement was the child’s phenomenological experience of parent’s resources available to them. And third, exposing and engaging the child in cognitive and intellectually stimulating activities was construed as cognitive/intellectual involvement. While the measures were specifically directed towards parent involvement in the educational context, the need to adapt the constructs, particularly personal involvement, and measures to the clinical context is obvious. Grolnick and Slowiaczek defined personal involvement in terms of the dedication of personal resources in the service of positive effect. We subscribe to the relevance of the definition. However, since clinical treatments usually address several issues that stretch beyond the scope of a single functional domain (e.g., reduce symptoms, function with friends and school, improve family relationships, etc.), is time limited and emotionally demanding, compared to schooling, the child may need the experience of parent’s emotional care beyond the context of the clinic. And a construct of parental care that is relatively stable across situations and time periods, for example, parental bonding [25] may be more appropriate.

Based on this literature review, the primary goal of this study was to examine the role of youth and parents’ report regarding the youths’ psychopathology and interpersonal problems on parent involvement in the treatment process. Further, because clinicians most often gather data from youth and their parents and discrepancy between youth and parents constitutes an important factor, we also examined the role of discrepancy on parent involvement.

Method

Subjects

Subjects were recruited from youth (ages 14–17 years) receiving outpatient treatment at the Stavanger University Hospital in Norway. A criterion for inclusion was that the youth had residence with at least one of the parents—biological or adoptive. There were 99 youth who completed the questionnaires. Of these, data were available from one or both parents for 66 subjects. Three subjects did not complete the whole set of questionnaires and were not included in the study. Data from the youth, mother, and father was available from 39 youth and data from youth and mother was available from all 63 youth. Due to the low response from fathers, we used only scores from the youth and mothers in this study. Further, although the analyses are primarily mothers’ report, we use the term parent in the rest of the article. The final cohort for this study comprised of 63 subjects. In total, 35% of the sample comprised of boys. All the subjects gave their informed consent and the study was approved by the regional ethics committee (REK-III), University of Bergen, Norway.

In total, 30 of the 63 subjects ended outpatient treatment and the average number of months in treatment was 12 (range = 24 SD = 6.63). Youth still in treatment averaged at 15 months (range = 19 SD = 6.58). Independent T-tests for continuous variables of age and number of months in treatment and chi-square tests for categorical variables of gender, diagnoses, and two-parent or single-parent families did not show any differences between youth who ended treatment or youth still in treatment. The sample consisted of 55.5% (n = 35) youth from dual-parent families and 45.5% (n = 28) from single-parent families. Mothers attended an average of 7.92 (median = 6 SD = 7.05) consultations and the youth attended an average 12.94 (median = 12 SD = 10.06) consultations.

Measures

Parent involvement variables

We used two measures of parent involvement based on the multidimensional constructs of parent involvement proposed and tested by Grolnick and Slowiaczek [9], namely, behavioral involvement (BI) and personal emotional involvement (PEI). First, parallel to their description of BI (e.g., attending school, meeting with the teacher, etc.) we used the number of consultations attended by the youth and the mother to construct BI. A similar approach to construing the frequency of consultations attended as a measure of family participation was recently reported by Hawley and Weisz [10]. However, the frequency of consultations parents and youth attend may be a function of the total consultations that the youth attended. Therefore, we used the percentage of the total number of consultations that the mothers attended as our measure of BI. PEI was measured using the Care sub-scale of the Parental Bonding Instrument—PBI [25]. A recent study with 1,071 Norwegian adolescents used the same scale as a measure of parent support and reported a Cronbachs’ alpha of 0.77 [23].

Child Behavior Checklist (CBCL) and Youth Self Report (YSR)

Youth and mothers described the mental health problems using the CBCL and the YSR [1]. CBCL is a parent report questionnaire that other important adults also can use to report the child’s (6–18 years) behavior and the latter is the YSR. Both questionnaires comprise of three scales assessing the child’s competence (Activities, Social, and School), eight narrowband syndrome scales (Anxious/Depressed; Withdrawn/Depressed; Somatic Complaints; Social Problems; Thought Problems; Attention Problems; Rule-Breaking Behavior and Aggressive Behavior), and two broadband scales of Internalizing and Externalizing disorders in addition to a total problem scale. The Norwegian version of the CBCL and YSR was first translated in 1986/1988 and later revised in 1993 and 2002. Nøvik [24] found that the Norwegian version of the CBCL had acceptable external and discriminant validity. However, because Norwegian norms are not available, we used the raw scores of the syndrome sub-scales in this study.

Inventory of Interpersonal Problems (IIP-C)

The Inventory of Interpersonal Problems was first developed by Horowitz et al., [13] to measure interpersonal complaints that people bring to therapy. Later, Alden, Wiggins, and Pincus [3] constructed circumplex scales (IIP-C) using 64 of the items from the original IIP. These self-report items are broadly divided into two types of statements. The first 39 items begin with “It is hard for me to” (for example) trust other people and keep things private from other people. The second 25 items start with “things I do too much” (for example: I fight with other people too much and I am too sensitive to criticism) Responses to each item are coded on a 5-point Likert scale (0 = not at all and 4 = extremely). There are eight scales that describe various themes: Domineering/Controlling (PA), Vindictive/Self-centered (BC), Cold/Distant (DE), Socially inhibited (FG), Nonassertive (HI), Exploitable (JK), Overly Nurturant (LM), and Intrusive/Needy (NO). Further, these scales are organized in a circumplex of interpersonal problems along the two dimensions of dominance and nurturance. The IIP-C has been shown to have good reliability. Alpha coefficients for the scales ranged between .72–.85 for the original North American sample [3] and .64–.85 in the current sample. This was not different from the range reported in two unpublished studies: one, with a Norwegian adult sample [15] reported a coefficient range of .69–.81 and two, the coefficient range with a Norwegian youth sample (14–18 years) was .69 and .83 [17].

Procedure and data analysis

Subjects included in the study had either received printed information regarding the research study prior to their first appointment or early on after their initial contact with the clinic. Those who agreed to participate gave their informed consent and completed the battery of measures either prior to the first consultation or shortly after commencement of assessment and treatment at the clinic. Parent and youth submitted the completed measure at the same time indicting no time lapse between their respective reports.

Data were examined by conducting three sets of multiple regression analyses with each of the dependent variables of BI and PEI—a total of six sets of linear regression analyses. All analyses were carried out using the statistical software, SPSS 13.0.

Results

The mean percentage of BI was 42.12 (SD = 25.28) and the mean score for PEI was 20.97 (SD = 3.16). Further, the mean scores and Pearson’s correlations between youth and mother reports for interpersonal problems and syndrome scales are presented in Table 1.

Table 1 Mean scores and Pearsons’ correlations between youth and parent report for syndrome scales of YSR and CBCL and the Inventory of Interpersonal Problems
Full size table

Highly correlated variables add little or nothing to a model in a regression analysis and therefore could lead to improper conclusions regarding their utility. There were low to medium correlations between youth and their mothers on the total sample. However, girls had more significant and higher levels of correlations with the mothers than did boys. However, the highest correlation between mothers and the girls was on externalizing disorders.

As can be seen in Table 2, the direct report of the youth and the parent was not significantly related to either BI or PEI. However, the discrepancy scores explained 47% of the variance for BI and the model was significant F = 2.52 (16) P < .01. Further, discrepancy scores explained 49% of the variance for PEI and the model was significant F = 2.76 (16) P < .01. This indicates that while direct reports of the youth and the parent, in and of themselves, were unrelated to parent involvement, the discrepancy scores were clearly related to parent involvement. Further, because discrepancy was measured at the beginning of the clinical contact and BI at the end of the study period, discrepancy can be said to predict BI.

Table 2 R2 and Beta values for individual Predictors of behavioral involvement (BI) and personal emotional involvement (PEI)
Full size table

Although BI and PEI are different sub-constructs of parent involvement, the overall pattern of relationship with direct and discrepant scores was noteworthy. While one may wonder if the two dependent variables are correlated, none of the 16 sub-scales were related to BI or PEI. Their differing pattern of association indicates that BI and PEI are separate sub-constructs of parent involvement. Table 2 also shows that twice the IIP-C scales have significant association with parent involvement than the syndrome scales of CBCL/YSR. When the syndrome scales entered the model as block 1 of the regression model, they were no significant relationships with either of the indices of parent involvement. However, when the IIP scales were added, the model was significant as noted above. This indicates that discrepancy scales of interpersonal problems were superior in predicting parent involvement.

Discussion

The primary focus of this study was to investigate the role of parent–youth discrepancy regarding the youth’s psychopathology and interpersonal problems on parent involvement in the assessment and treatment of youth. It was proposed that discrepancy exerts its effect by diminishing parent involvement in central socializing and help-seeking aspects of the youth’s life; of particular interest to this study was parent involvement in assessment and treatment of youth referred to specialty mental health clinics. A related focus was to investigate if there was a difference between the predictive value of parent–youth discrepancy on two domains of problems, namely symptoms/syndromes and interpersonal problems. The main results showed that while direct report of youth and parent were not successful predictors, discrepancy scores successfully predicted both the parent involvement variables. And although BI and PEI are two different aspects of parent involvement, their overall similarity of associations with the explanatory factors indicates they are indeed related as indices of parent involvement.

Ideally, the parent’s direct report should have predicted the level of their involvement for at least two reasons: first, direct report could be construed as the informant’s felt distress on a given measure. And therefore, should have been adequate in predicting the level of therapeutic attention—the higher the distress levels the more consultations. Second, because parents are key players in identifying the problems, initiating contact and providing clinical information to therapists’ [30] their involvement should have been high. However, the lack of association between the direct report of the youth and parents with parent involvement may have a developmental explanation. It could be argued that given the increased need for autonomy in adolescence, parents restricted their BI in the treatment process, and instead, provided emotional support at home. However, the lack of significant relationship between PEI and direct report does not support this notion. Yet another explanation may be that the phase of the treatment may have had an effect on parent involvement. A previous study showed most contact between clinicians and parents in the intervention phase of the treatment [16].

There are two obvious possibilities for the association between high discrepancy scores and low parent involvement. First, parents may not wish to be involved in the treatment process and second, the youth may not want to involve their parents. When parents do not wish to be involved, it may be because they do not understand why the youth requires treatment or why they should be involved. Additionally, relationship factors between parents and youth, communication problems and other factors, such as, the presence of parental psychopathology may be among the common explanations for why parents may not want to be involved in the treatment process. Occasions where youth do not want their parents involved may be associated with problems that the youth wishes to conceal from the parents—problems they know are unacceptable. For example, problems with drug and alcohol misuse and other conduct problems. Discrepancy on behavioral problems predicted behavioral problems in young adulthood in the Ferdinand et al. study. They also found a similar trend for discrepancy regarding emotional problems and persistence of emotional problems in young adulthood.

Previous studies have identified youth as better reporters than their parents on internalizing problems [6, 20] and interpersonal problems [21]. Consequently, it may be argued that when parents fail to correctly identify youth’s interpersonal problems, and the youth are able to better report these problems, the need for comprehensive parent involvement may be unnecessary. Individual scales that were statistically significant and related to BI, were scales that are associated with individuals who typically are compliant, avoid conflict, do not readily make known their distress and problems to others and have responded well to psychotherapy [12]. However, scales associated with PEI were those that are typically related to externalizing disorders. This implies that youth reporting more externalizing problems may have trouble experiencing their parents as emotionally involved and caring. This is not surprising given that parents may be strict with these youth. However, irrespective of the type of problem, without proper guidance and help parents may find it difficult to balance between adequate monitoring and autonomy granting. Clinicians may have a valuable contribution to make by enhancing communication and reducing relational problems between youth and parents.

Conclusions of this study must be tentative due to several limitations. A total of 48 variables were studied for each of the two dimensions of parent involvement. The study was conducted with a relatively small sample size and this combination runs the risk of type II error and limits the ability to generalize to the population. Further, the small sample size limited the scope of the study that could have included examining other relevant research issues like, the direction of discrepancy (e.g., youth reporting more problems than parents and vice versa). The low response rate from fathers has limited the scope of the study, especially in the light of research showing relationship between father–youth psychopathology [26].

Yet another limitation was the mixed use of cross-sectional and longitudinal structure. All the self-report measures including PEI were completed at the beginning of the clinical contact while only BI was measured at the end of the study period with an average of 12 consultations for subjects who ended treatment and an average of 15 consultations for subjects still in treatment. Thus the longitudinal structure related to BI provides support for positing discrepancy scores as predictors of BI. However, PEI is a stable trait-like construct and may be unlikely to change over this relatively short study period. A recent study demonstrated the stability of the PBI over a 20-year period [29]. However, because PEI lacks the longitudinal structure, results regarding its predictive index should be treated cautiously. To our knowledge, this is the first study with a referred clinical population that has examined the role of parent–youth discrepancy on parent involvement in outpatient treatment. We acknowledge the need for additional research with a larger, representative group with a longitudinal design to verify the results of this study.

The primary focus of this study was to examine the role of the youth and parents’ report regarding the youths’ psychopathology and interpersonal problems and discrepancy between their accounts on parent involvement. The main result of this study was that direct reports of the youth and parent were not related to parent involvement, but discrepancy scores between parents and youth were significantly related to parent involvement.