Publication - Research and analysis

Factors affecting children’s mental health and wellbeing: findings

Published: 8 Jan 2020

Results from the 2015- 2017 Realigning Children's Services Wellbeing Surveys into factors effecting mental health and wellbeing amongst children and young people in Scotland.

56 page PDF

725.8 kB

56 page PDF

725.8 kB

Contents
Factors affecting children’s mental health and wellbeing: findings
2 Methods

56 page PDF

725.8 kB

2 Methods

This report uses data from the RCS Children’s Wellbeing Surveys in Scottish primary and secondary schools. It examines what risk and protective factors are associated with emotional and behavioural problems and positive mental wellbeing.

2.1 The RCS Children’s Wellbeing Surveys

As part of the Realigning Children’s Services (RCS) programme, pupils in participating local authorities completed the school-based Children’s Wellbeing Surveys. These census-level surveys gathered data from children and young people, who agreed to participate, on their subjective health and wellbeing. The survey responses were securely linked to local administrative and geographical data about the respondents.

This report uses available data from the primary (P5-P7) and secondary (S1-S4) school-based surveys in five local authorities that participated in the RCS programme between 2015 and 2017. In total, survey responses and administrative data were available for 32,154 secondary pupils and 24,797 primary pupils from these five local authorities.

The report analyses the primary and secondary school surveys separately. The primary school survey included fewer survey questions, with simplified wording and response options. However, both surveys broadly capture similar subjective experiences, with age-appropriate measures of poor mental health, positive mental wellbeing and relevant factors in other domains of life.

2.2 Measures of mental health and wellbeing

2.2.1 Poor mental health

In the secondary survey, poor mental health was measured using the Strengths and Difficulties Questionnaire (SDQ)[7]. Pupils’ scores on two subscales provide a measure of two important components of child and adolescent mental health problems: emotional problems and conduct problems. This report examines prevalence rates and associated risk factors for ‘very high’ scores on each of these subscales. Using the ‘very high’ established cut-off score identifies children with the most severe problems in these areas[8].

In the primary survey, pupils rated how often they experienced six different positive and negative emotions. Their answers were combined into an overall score of positive mood[9]. This report examines factors associated with having a lower than average mood score. Children in this group tended to experience negative feelings more often (and positive feelings less often) than other children in the survey sample.

2.2.2 Positive mental wellbeing

Positive mental wellbeing captures more than simply an absence of mental health problems[10]. It encompasses subjective experiences (e.g. happiness and life satisfaction) and psychological functioning (e.g. confidence, energy and purpose)[11].

In the secondary survey, positive mental wellbeing was measured using the Warwick-Edinburgh Mental Wellbeing Scale (WEMWBS)[12]. The WEMWBS provides an overall score and was not developed to include established cut-offs for high or low wellbeing. The RCS wellbeing surveys use a cut-off score defined as one standard deviation above the mean WEMWBS score in the 2015 Scottish Schools Adolescent Lifestyle and Substance Use Survey (SALSUS) survey to identify pupils with above average positive mental wellbeing[13].

In the primary survey, pupils completed a measure of life satisfaction, which is one component of positive mental wellbeing[14]. Pupils answered five adapted questions from the Student’s Life Satisfaction Scale[15], which were combined to give an overall score[16]. The analysis examined factors associated with having a higher than average life satisfaction score, compared to other children in the survey sample.

2.3 Potential risk and protective factors

The analysis examined whether these outcome measures were associated with potential risk or protective factors across broad domains of children’s lives. These domains are:

  • Family, including quality of parent-child relationships and family time
  • School, including teacher-child relationships and enjoyment of school
  • Peer, including relationships with friends and experience of bullying
  • Area, including perceptions of local area safety and availability of outdoor space
  • Health, including subjective general health and physical activity

Existing research supports that these aspects of children’s lives are important for wellbeing and mental health, and the RCS school-based surveys were designed to capture relevant factors associated with wellbeing. Full lists of all risk factors examined in each domain are presented in Table 1 for secondary school pupils (Section 3.2) and in Table 5 for primary school pupils (Section 4.2).

Since this report examines both positive and negative outcomes, depending on the context, the report refers to the same factors (e.g. attitude towards school) as either ‘risk factors’ for negative outcomes or ‘protective factors’ for positive outcomes. For example, in the case of attitude towards school, it examines whether not liking school is a ‘risk factor’ for emotional problems, and whether liking school is a ‘protective factor’ for positive mental wellbeing. The categories are the same in both cases: inverting the comparison simply makes results more intuitive to interpret.

2.4 Analysis approach

The analysis in this report considered the primary and secondary school surveys separately, since they used different survey questions. To answer the research questions outlined in Section 1.3, the analysis involved the following steps.

For each survey (primary and secondary), the analysis first calculated survey-weighted percentages to examine:

1. How prevalent each mental health outcome was (overall and by school year, gender and area deprivation)

2. How prevalent each potential risk factor was

Then, for each mental health outcome in turn, binomial logistic regression models examined:

3. Whether a child’s chances of having that mental health outcome were associated with:

a. Having each individual risk factor

b. Having multiple risk factors in a domain[17]

4. Whether these associations remained significant after accounting for the combined influence of other factors simultaneously

All analysis was completed using R[18]. All associations presented in the report are statistically significant (p<0.05), unless stated otherwise. All associations controlled for relevant child characteristics, outlined below.

2.4.1 Control variables

All associations that were examined accounted for a number of control variables to allow meaningful comparisons between similar children with or without a certain risk factor.

For primary surveys, the analysis controlled for: school year, gender, area deprivation (Scottish Index of Multiple Deprivation; SIMD) and free school meal eligibility.

For secondary surveys, additional measures were available and the analysis controlled for: school year, gender, area deprivation and free school meal eligibility (as above); plus child ethnicity and household composition (two parent, single parent, step parent or other families, and number of siblings).

2.4.2 Missing data

There was some missing data, as some children skipped certain survey questions. The analysis therefore used all available data on each measure when presenting prevalence rates, but limited the remaining analysis stages to pupils who had completed the outcome measures (see Section 2.2). This gave analysis samples of 22,935 secondary school pupils and 20,989 primary school pupils.

Within these analysis samples, 26% of primary pupils and 47% of secondary pupils had skipped at least one question of interest in the current analysis. Children who skipped questions were more likely to be younger, live in a deprived area, and report poorer relationships with family and peers, and poorer perceptions of their school and neighbourhood environments. Therefore simply removing children with incomplete data would give a biased sample, so the analysis used multiple imputation (a robust statistical method for dealing with missing data) [19].

2.4.3 Survey weights

All analyses used survey weights from the RCS wellbeing survey datasets. Weights were computed based on gender, school year and school denomination (non-denominational / Roman Catholic) for secondary school pupils; and gender, school year and free school meal eligibility for primary school pupils. This corrects for any over- or under-representation of these characteristics in the samples, to bring the sample profile in line with the population profile of P5-P7 and S1-S4 pupils in these five local authorities.

2.5 Strengths & limitations of this analysis

Strengths of this analysis approach include: using established, validated questionnaire measures of mental health and wellbeing; controlling for child characteristics; accounting for the simultaneous influence of other factors in the final analysis step; and using multiple imputation of missing data to reduce bias and improve power.

However, this analysis has a number of limitations. The RCS wellbeing surveys are cross-sectional: each participating pupil answered all survey questions at one time point. Therefore, although this analysis can identify an association between a particular risk factor and mental health outcome, it cannot establish the direction and temporality of this association or whether the risk factor causes poorer mental health outcomes. For example family conflict may lead to emotional problems, but emotional problems may also lead to family conflict.

The list of factors examined in this report is not intended to be comprehensive: there are likely to be additional relevant factors that influence mental health and wellbeing, beyond the current dataset or analysis. Furthermore, since the primary and secondary school surveys used different (age-appropriate) questions and response options, it is not possible to directly compare the results across age groups.


Contact

Email: franca.macleod@gov.scot