Perceived need and barriers to adolescent mental health care: agreement between adolescents and their parents

Aims Mental disorders cause high burden in adolescents, but adolescents often underutilise potentially beneficial treatments. Perceived need for and barriers to care may influence whether adolescents utilise services and which treatments they receive. Adolescents and parents are stakeholders in adolescent mental health care, but their perceptions regarding need for and barriers to care might differ. Understanding patterns of adolescent-parent agreement might help identify gaps in adolescent mental health care. Methods A nationally representative sample of Australian adolescents aged 13–17 and their parents (N = 2310), recruited between 2013–2014, were asked about perceived need for four types of adolescent mental health care (counselling, medication, information and skill training) and barriers to care. Perceived need was categorised as fully met, partially met, unmet, or no need. Cohen's kappa was used to assess adolescent-parent agreement. Multinomial logistic regressions were used to model variables associated with patterns of agreement. Results Almost half (46.5% (s.e. = 1.21)) of either adolescents or parents reported a perceived need for any type of care. For both groups, perceived need was greatest for counselling and lowest for medication. Identified needs were fully met for a third of adolescents. Adolescent-parent agreement on perceived need was fair (kappa = 0.25 (s.e. = 0.01)), but poor regarding the extent to which needs were met (kappa = −0.10 (s.e. = 0.02)). The lack of parental knowledge about adolescents' feelings was positively associated with adolescent-parent agreement that needs were partially met or unmet and disagreement about perceived need, compared to agreement that needs were fully met (relative risk ratio (RRR) = 1.91 (95% CI = 1.19–3.04) to RRR = 4.69 (95% CI = 2.38–9.28)). Having a probable disorder was positively associated with adolescent-parent agreement that needs were partially met or unmet (RRR = 2.86 (95% CI = 1.46–5.61)), and negatively with adolescent-parent disagreement on perceived need (RRR = 0.50 (95% CI = 0.30–0.82)). Adolescents reported most frequently attitudinal barriers to care (e.g. self-reliance: 55.1% (s.e. = 2.39)); parents most frequently reported that their child refused help (38.7% (s.e. = 2.69)). Adolescent-parent agreement was poor for attitudinal (kappa = −0.03 (s.e. = 0.06)) and slight for structural barriers (kappa = 0.02 (s.e. = 0.09)). Conclusions There are gaps in the extent to which adolescent mental health care is meeting the needs of adolescents and their parents. It seems important to align adolescents' and parents' needs at the beginning and throughout treatment and to improve communication between adolescents and their parents. Both might provide opportunities to increase the likelihood that needs will be fully met. Campaigns directed towards adolescents and parents need to address different barriers to care. For adolescents, attitudinal barriers such as stigma and mental health literacy require attention.

C C C E Note N/n: unweighted number of respondents with 'N' representing total population and 'n' representing subpopulation.
The categories of the dependent variables are (see also table below A, B, C, D): • A: Adolescents and parents agree on fully met need (reference group, n=131), • B: Adolescents and parents agree on either partially met or unmet need (n=93), • C: Adolescents perceived a greater level of need than their parents (n=488), • D: Parents perceived a greater level of need than their child (n=407), • E: Adolescents and parents agree on having no need (n=1,191).
eMaterial 1 Additional measures, how they were used in this study. What was measured Measure (questionnaire or questions) How we used the measure (categories of measures are indicated in italic)

Regression used as: Adolescents' probable disorder
Sum score of four subscales of the 'Strengths and Difficulties Questionnaire' (SDQ) (Goodman 1997(Goodman , 2001 for total difficulties: emotional symptoms, hyperactivity, conduct problems, and peer problems. Total difficulties score ranges from 0-40 with higher scores indicating more problems and with 80% of adolescents in the community having normal levels of difficulties and 10% either borderline or abnormal levels. Abnormal levels are referred to as an indication for having a probable disorder (Goodman 2001). For total difficulties in this study of adolescents aged 13-17: Cronbach's (parent report)=0.72; Cronbach's (adolescent report)=0.70.
We defined the presence of a probable disorder as either adolescents or parents reporting total difficulties on an abnormal level, while we defined the absence of probable disorder as both adolescents and parents reporting total difficulties on a normal or borderline level. This allowed us to consider both adolescents' and parents' perceptions about adolescent mental health problems because their agreement on symptoms is typically modest (Rescorla et al. 2013).

Independent variable
Adolescents' probable internalising disorder or probable externalising disorder SDQ subscales can be summarised into internalising (emotional and peer subscale) and externalising (conduct and hyperactivity subscales) problems (Goodman et al. 2010). Each of the subscale ranges from 0-10 with higher scores indicating more problems and with 80% of adolescents in the community having normal levels of problems in respective subscale and 10% either borderline or abnormal levels. Abnormal levels are used as an indication for having a probable emotional, conduct or hyperactivity disorder (Goodman 2001). For probable internalising disorder in this study: Cronbach's (parent report)=0.62; Cronbach's (adolescent report)=0.60. For probable externalising disorder in this study Cronbach's (parent report)=0.70; Cronbach's (adolescent report)=0.68.
First, separately for adolescents and parents, probable internalising disorders were defined as reporting abnormal levels on the subscales emotional and/or peer problems; probable externalising disorders as reporting abnormal levels on conduct and/or hyperactivity problems. Second, we defined the presence of a probable internalising disorder as either adolescents or parents reporting it on an abnormal level, while we defined the absence of probable internalising disorder as both adolescents and parents reporting internalising disorders on a normal or borderline level. We proceeded accordingly for the presence and absence of externalising disorders. This was done before by (Downs et al. 2013 We collapsed four categories into two ('a lot/some' and 'a little/not at all').

Independent variable
General family functioning Reliable and valid six items version (Boterhoven De Haan et al. 1985) of the McMaster Family Functioning scale (Epstein et al. 1983). Items are rated on a 4-point likert scale, answers are summed up and divided by the number of items to receive a score between 1-6 (Miller et al. 1985). In this study including parents with adolescents aged 13-17, Cronbach's =0.87.
Used as proposed by Miller and colleagues (1985): score of >2 indicate healthy and 2 unhealthy family functioning.

Control variable
Parental psychopathology (1) Parents levels of psychological distress in the past four weeks was assessed with the 10-item Kessler Psychological Distress Scale (Kessler et al. 2003). In this study including parents with adolescents aged 13-17, Cronbach's =0.90.
(2) Question for lifetime disorder: 'Have you ever been told by a doctor or mental health professional that you have any of these problems?': • Panic attacks Based on an earlier YMM study (Johnson et al. 2018), parental psychopathology was considered to be present if parents reported a lifetime diagnosis of a mental disorder and/or (very) high levels of psychological distress in the past four weeks according to the 10-item Kessler Psychological Distress Scale. Otherwise, parental psychopathology was considered to be absent. We collapsed these four categories into two: (1) family with two biological parents or (2) other family type.

Control variable
Remoteness YMM assessed remoteness according to the Australian Statistical Geography Standard-Remoteness Area (ASGS-RA) provided by the ABS. It is a geographical classification which defines place of residence/location in terms of remoteness. Remoteness is categorised as: (1) Major cities of Australia, (2) inner regional Australia, (3) outer regional Australia, (4) remote Australia.

Control variable
Socio-economic advantage and disadvantage (IRSAD) YMM assessed socio-economic advantage and disadvantage according to the Socio-Economic Index for Areas (SEIFA index) of the ABS. The SEIFA index defines socio-economic advantage and disadvantage according to the place of residence rather than to the individuals' actual status of socio-economic advantage and disadvantage. The SEIFA index contains the 'index of relative socio-economic advantage and disadvantage' (IRSAD) which was used in this study. IRSAD can be divided into quintiles with lowest quintile (most disadvantaged) to second quintile, to third quintile, to fourth quintile and to highest quintile (most advantaged).
We collapsed IRSAD quintiles into advantaged (highest and fourth quintile) and disadvantaged (lowest, second and third quintile). Reference categories of independent/control variables are: male sex, advantaged IRSAD, major cities (remoteness), living in another family constellation (family type), bachelor degree or higher (parental education), normal family functioning, parent has 'a lot/some' knowledge of adolescent feelings. The presence of a probable disorder was defined as either adolescents or parents reporting total difficulties (SDQ) on an abnormal level. Note RRR: relative risk ratio; CI: confidence interval; F(24, 525)=2.43. eTable 8 All results of multinomial logistic regressions of variables associated with patterns of agreement on perceived need; internalising and externalising disorder separately (n=1,119).

16
3. Not all adolescents participated in self report. Variable yint11 was used to identify responders and delete all non-responders from data set.

Perceived Need
Plot for overall perceived need for care (any type of care) and for perceived need for four different types of care (information, medication, counselling, skill-training) Same plot as above but only including subsample of adolescents with a mental health problem.

Agreement on overall perceived need
Circos https://jokergoo.github.io/circlize_book/book/the-chorddiagram-function.html to plot agreement and disagreement of adolescents and parents on overall perceived need (for any type of care

Overall Perceived Need
Barriers to care Next, we will have a look at the distribution of adolescents and parents barriers to care (first without then second with survey weighted including confidence intervals).

Couldn't get appointment (S)
Couldn't afford it (S) Problem finding service that could help (S)