Universal and targeted computerised cognitive behavioural therapy (Think, Feel, Do) for emotional health in schools: results from two exploratory studies



Background:  Computerised Cognitive Behaviour Therapy (cCBT) offers the potential to make evidence based interventions more available and accessible for children and young people. Research is currently limited and proof of concept studies are required to explore the viability and possible therapeutic benefits of cCBT as a universal or targeted intervention in schools.

Method:  In Study 1, cCBT was provided as a universal intervention to 13 participants who were assigned to either group cCBT or a matched computer gaming condition. In Study 2, cCBT was provided as a targeted intervention by trained school nurses to 12 participants with mild or moderate emotional problems.

Results:  Both studies found cCBT to result in immediate post intervention benefits and feedback about the programme was very positive.

Conclusion:  This study provides proof of concept that cCBT provided as a universal or targeted emotional health intervention in schools is viable and may result in immediate therapeutic benefits. These findings are limited by the small sample size and absence of follow-up but suggest that more methodologically robust evaluations should be pursued.

Key Practitioner Message:

  •  Anxiety and low mood are common in children and adolescents and can cause significant impairment in functioning
  •  Access to specialist mental health services for young people is limited
  •  cCBT programmes offer the potential to increase access to treatment
  •  cCBT delivered as a school-based emotional health intervention may reduce emotional symptoms in children and adolescents


Emotional problems of anxiety and low mood are common in children and adolescents and can cause significant impairment in social and academic functioning (Ford, Goodman, & Meltzer, 2003a; Weems & Stickle, 2005). Studies suggest that Cognitive Behaviour Therapy (CBT) is effective in both the treatment and prevention of depression and anxiety disorders in young people (Compton et al., 2004; Klein, Jacobs, & Reinecke, 2007; Shochet et al., 2001; Soler & Weatherall, 2007; Stallard et al., 2008). However, comparatively few young people with emotional problems benefit from CBT. Few have contact with specialist mental health services (Ford, Goodman, & Meltzer, 2003b), and there are a limited number of appropriately trained child CBT specialists (Stallard et al., 2007). This suggests a need to explore how potentially effective interventions can be made more available and accessible to those who might benefit from them.

In terms of accessibility, there has been recent interest in providing mental health interventions in schools. In the UK, the Department of Education and Skills has implemented major initiatives such as SEAL (Social and Emotional Aspects of Learning) and TAMHS (Targeted Mental Health in Schools) designed to promote the emotional skills thought to underpin effective learning and positive behaviour (DCSF, 2008; DFES, 2005). School based interventions have typically been provided in two main ways; universally to all children in order to promote emotional resilience and prevent the development of mental health disorders or in a targeted way, focusing upon those at increased risk of developing mental health problems or already displaying mild/moderate problems in order to reduce or prevent symptoms from worsening.

Increasing the availability of specialist interventions has resulted in the use of technology and the development of computerised interventions, particularly computerised cognitive behaviour therapy (cCBT). A number of cCBT programmes have been developed and evaluated with adults, leading the National Institute of Clinical Excellence in the UK to recommend cCBT as a first step treatment for anxiety and depression in adults (NICE, 2006).

Research into the use of cCBT for the treatment of anxiety and depression in children and adolescents is however in its infancy (See Richardson, Stallard, & Velleman, 2010, for a review). March, Spence and Donovan (2008) and Spence et al. (2006) conducted randomised controlled trials of the program, ‘BRAVE online’, comparing cCBT (= 40; = 27 respectively) to a wait list group (= 33; = 23) and found significant improvements in diagnostic status, functioning and behaviour. Case series of ‘Cool teens’ and ‘BRAVE online’ have also found promising results (Cunningham et al., 2009; Spence et al., 2008), as has a case series of a cCBT programme called ‘Stressbusters’ for adolescent depression (Abeles et al., 2009).

In terms of attitudes, parents tend to be positive towards the use of cCBT with children and can identify a number of benefits (Stallard, Velleman, & Richardson, 2010). However, child mental health professionals consider cCBT to be more effective as a supported intervention for mild and moderate problems rather than as a stand alone intervention for more significant disorders (Stallard, Richardson, & Velleman, 2010).

This paper reports on two proof of concept studies to explore the viability and possible immediate therapeutic benefits of one cCBT programme, ‘Think, Feel, Do’, delivered in schools as a universal and targeted emotional health intervention.

Method: Study 1 - Universal intervention

Design and procedure

Ethical approval was granted by the Graduate School of Education at the University of Bristol. An opportunistic sample of primary school boys was recruited to explore the effect of ‘Think, Feel, Do’ as a universal group intervention on symptoms of anxiety. Parents and children were required to provide signed consent and the boys were assigned to either cCBT or a matched computer time (gaming) condition. Both conditions received six sessions of their assigned programme, which were completed during normal class time. Each child had their own computer and a researcher (MA) was present at all times to guide the children through the program and answer any questions.


Participants were all boys between the ages of 10 and 12 years (= 10.6, SD = 0.65) who were recruited from two co-educational primary schools in Bristol. An opportunistic sample of 22 primary school boys were recruited to take part in the study and were randomly assigned to either the cCBT or computer gaming condition. Complete parent and child pre- and post-assessment measures were obtained from 13 boys and their parents who consented to take part in the study; of these, 6 boys were assigned to the group cCBT condition and 7 to the gaming group. The data from these children are presented in this paper.


Parents completed self-report measures at home, and children at school, before the first session and immediately after completing the programme. The following standardised measures were used:

Spence Children’s Anxiety Scale Child and Parent Version (SCAS; Spence, 1998).  These child- and parent-report scales measure anxiety in children aged 8 to 12 years. Both comprise six subscales: separation anxiety, social phobia, obsessive compulsive disorder, panic/agoraphobia, generalised anxiety, and physical injury fears. The scale has high internal consistency (Cronbach alpha = .93) and a 6-month test-retest reliability of 0.6 (Spence, 1998).

Strengths and Difficulties Questionnaire Parent Version (SDQ; Goodman, 1997).  The parent-rated behavioural screening questionnaire is for children and adolescents aged 3 to 16 years. It consists of 25 items, with subscales measuring total difficulties, emotional problems, conduct problems, hyperactivity, peer difficulties, and prosocial behaviour. The scale has good internal consistency (mean Cronbach alpha = .73) and a 4–6-month test-retest reliability of 0.62 (Goodman, 2001).

A series of brief interview questions were designed by one of the authors (MA) to assess perceptions of the cCBT programme. All of those in the cCBT condition were invited to take part in a focus group to discuss the programme. Eight boys from the cCBT condition took part in two interviews (four boys in each), which were conducted by the researcher (MA) and held at the children’s school on completion of the cCBT programme.


The cCBT group used the programme ‘Think, Feel, Do’, an interactive, multimedia CD-ROM based on the CBT manual Think Good-Feel Good (Stallard, 2002). ‘Think, Feel, Do’ is designed for emotional problems of anxiety and mood and is facilitated by a teacher, nurse or assistant psychologist. The programme consists of six, 45-minute sessions and covers themes that include emotion recognition and management; linking thoughts, feelings and behaviour; identifying and challenging negative thoughts; and problem solving. The programme is based on CBT principles and has a psychoeducational element, with cartoon characters guiding users through the various activities. It is highly interactive and involves quizzes, practical exercises, video clips, music and animation. A more detailed description is provided elsewhere (McCusker, 2008; Stallard et al., 2011). The cCBT group had weekly 45-minute sessions with the researcher. The computer gaming group had six matched 45-minute sessions with the researcher but instead played online games. This ensured that children had a similar amount of computer time and controlled for the time the children spent out of their normal classroom routine and the time spent with the researcher.

Data analysis

For this study the sample size was small, and exploratory analysis revealed that the quantitative data were not normally distributed. The results were therefore subjected to non-parametric paired analyses for each condition separately. Between-group analysis was not conducted as a result of insufficient sample size.

Method: Study 2 - Targeted Intervention

Design and procedure

Ethical approval was obtained through a National Health Service Ethics Committee as part of a larger trial. A case series was used and examined the pre- and post-intervention scores of children and adolescents identified by their school as having emotional health problems. The sample was opportunistic, with potential children being approached by their school nurse and child and parent consent obtained. The intervention, ‘Think, Feel, Do’, was delivered on a one-to-one basis by the school nurse over 6 sessions at school. These were usually delivered weekly, sometimes every 2 weeks, although the average timescale of treatment is unknown as data were not collected on this. Assessments were completed before and after the intervention by a researcher (TR) who was not involved in recruitment or the delivery of the intervention. Assessments were usually completed within 2 weeks of the intervention being completed, although again the exact time scale is not known.


Participants were recruited by school nurses from eight schools in Bath and the surrounding area. Participants were invited to take part if they had been identified by teachers or the school nurse as having mild or moderate emotional problems of anxiety or low mood. Because of the exploratory nature of this study formal records of the number of children approached and rates of attrition were not kept. The sample described in this paper consists of 12 children, 9 boys and 3 girls aged 10 to 16 years, with a mean of 12.4 years (SD = 1.53).


Self-report measures were used for this study, as assessments were usually undertaken at school so parents were not present. Participants completed the following standardised measures:

Spence Children’s Anxiety Scale Child and Parent Version (SCAS; Spence, 1998).  See description above.

Adolescent Well-Being Scale (AWS; Birleson, 1980).  This measures depression in children and adolescents between the ages of 11 and 16 years. The scale has 18 items responded to as ‘Most of the time’, ‘Sometimes’ or ‘Never’. There are no subscales; a total score above 12 indicates a depressive disorder. This scale has good internal consistency (Cronbach alpha = 0.88, Ivarsson & Gillberg, 1997).

Schema Questionnaire for Children (SQC; Stallard & Rayner, 2005).  This measures negative thinking patterns/schemas in children and adolescents. There are 15 items that are rated from 1 to 10, based on how strongly they are believed. An overall score is produced indicative of total negative thinking. There are no subscales or cut-off points for this scale. This scale has good internal consistency (Cronbach alpha = 0.82, Stallard & Rayner, 2005).

Rosenberg Self Esteem Inventory (RSEI; Rosenberg, 1965).  This measures global self-esteem with 10 items that are responded to as ‘Strongly agree’, ‘Agree’, ‘Disagree’ or ‘Strongly disagree’. An increasing total score represents higher self-esteem, and scores lower than 15 are suggestive of low self-esteem. This scale has good internal consistency (Cronbach alpha = 0.81, Schmitt & Allik, 2005).

Satisfaction questionnaire.  An author constructed questionnaire was designed to survey satisfaction with the intervention, consisting of 7 items rated on Likert scales of 1–9.

Data analysis

There were no missing data for participants who completed pre and post measures. Exploratory analyses revealed a normal distribution for data from this study. However, due to the small sample size, results were analysed using non-parametric paired sample tests.

Results: Study 1 - Universal intervention

Pre-intervention comparison

Mann-Whitney non-parametric analyses revealed no significant pre-intervention differences between the cCBT and gaming groups on total child or parent-rated anxiety. The total Strengths and Difficulties Questionnaire was approaching significance (U = 3, = .052), with children in the gaming group being rated more highly on this measure.

Intervention effects

Table 1 displays the results of the one-tailed Wilcoxon Signed-Rank tests. These revealed a statistically significant post-intervention reduction in total child anxiety ratings on the SCAS-C for the cCBT (= −1.79, < .05), but not for the gaming condition (= −1.59, > .05). The mean total anxiety score on the SCAS-C for a community sample of boys within the 8–12-year age group is reported as 27.2 (SD = 16.15) (Spence, 1998). The cCBT group had a higher total anxiety score than the gaming group at both pre-intervention (M = 29.0, SD = 18.5; = 18.4, SD = 11.5 respectively) and post-intervention (= 21.3, SD = 17.5; = 14.29, SD = 12.2 respectively). Analysis of subscale scores revealed a statistically significant post-intervention reduction on the social anxiety (= −1.80, < .05) and generalized anxiety (= −1.73, < .05) sub-scale scores for the cCBT group only. There was no significant post-intervention change in parent-rated SDQ or SCAS scores, although reductions on the total SCAS score in the cCBT condition were approaching significance (= −1.59, = .056). The mean total difficulties score for the parent-rated SDQ has been identified for a community sample of males between 5–15 years (= 9.1, SD = 6.0) (Meltzer et al., 2000). The gaming group had a higher total difficulties score than the cCBT group at both pre-intervention (= 8.3, SD = 4.1; = 3.5, SD = 2.6 respectively) and post-intervention (= 8.3, SD = 4.3; = 4.0, SD = 2.7 respectively). The mean total anxiety score on the parent-rated SCAS has been reported for a community sample of boys between 6–11 years (= 16.0, SD = 11.6) (Nauta et al., 2004). The mean total anxiety score for cCBT and gaming groups were similar at both pre-intervention (= 13.5, SD = 11.1; = 14.1, SD = 7.8 respectively) and post-intervention (= 10.5, SD = 11.1; = 12.6, SD = 9.9).

Table 1.  Pre and post scores for participants in cCBT and computer gaming group in study 1 (= 13)
SubscalecCBT groupPlacebo group
Pre mean (SD)Post mean (SD) Z scorePre mean (SD)Post mean (SD) Z score
  1. *< .05

SCAS – Child
 Total29.0 (18.5)21.3 (17.5)−1.787*18.4 (11.5)14.29 (12.2)−1.59
 Separation anxiety4.5 (3.7)3.5 (3.6)−1.3422.4 (2.9)1.4 (0.9)−1.062
 Social anxiety6.7 (4.9)4.8 (5.6)−1.802*3.4 (3.6)2.7 (3.3)−0.736
 Obsessive compulsive4.3 (3.0)3.5 (2.9)−1.5184.3 (4.2)2.3 (2.9)−1.461
 Panic/agoraphobia2.5 (1.9)1.2 (2.9)−0.9623.1 (2.9)2.7 (3.9)−0.422
 Physical injury fears4.7 (4.6)4.5 (4.8)−0.1842.3 (2.1)2.4 (1.5)−0.171
 Generalised anxiety6.3 (2.9)3.8 (2.5)−1.725*2.9 (1.7)2.7 (2.7)−1.414
SCAS – Parent
 Total13.5 (11.1)10.5 (11.1)−1.59014.1 (7.8)12.6 (9.9)−1.051
 Separation anxiety2.3 (1.9)1.8 (1.7)−1.0894.0 (3.7)3.4 (3.4)−1.414
 Social anxiety3.5 (2.6)2.17 (2.4)−1.5112.7 (2.2)2.7 (3.5)0
 Obsessive compulsive2.0 (2.8)2.2 (2.7)−0.4471.6 (1.4)1.3 (1.4)−0.412
 Panic/agoraphobia1.0 (2.0)1.0 (2.0)00.7 (1.5)0.9 (1.1)−0.368
 Physical injury fears2.0 (2.4)1.0 (1.1)−1.292.4 (2.9)2.4 (2.2)0
 Generalised anxiety2.7 (2.3)2.3 (2.3)−12.6 (1.1)2.0 (1.8)−0.921
 Total difficulties3.5 (2.6)4.0 (2.7)08.2 (4.1)8.3 (4.3)0
 Emotional difficulties0.0 (0.0)0.6 (0.9)−11.0 (0.9)0.9 (1.2)0
 Conduct problems0.8 (0.5)0.6 (0.5)−11.7 (0.8)1.4 (1.1)−1.134
 Hyperactivity2.2 (2.1)2.4 (1.7)−0.5573.5 (1.9)4.6 (2.3)−0.73
 Peer problems0.5 (0.6)0.4 (0.5)−12.0 (2.6)1.4 (1.3)−0.378
 Prosocial9.0 (2.0)9.4 (0.5)−0.4478.3 (2.3)7.4 (2.2)−1

Participant satisfaction

Participants were generally positive about ‘Think, Feel, Do’ and no sessions were identified as unhelpful. The favourite sessions were those that covered thinking traps (negative thoughts), the body signals associated with different emotions, and relaxation techniques. The majority enjoyed the appearance of the programme and reported that they had used some of the skills to deal with problems or cope with negative emotions. On the negative side, some of the younger children found some of the concepts and ideas challenging.

Results: Study 2- Targeted intervention

Intervention effects

Table 2 displays the results of Wilcoxon Signed-Rank t-tests for Study 2. These showed that there were statistically significant improvements post-intervention on the AWS Total Score (Depression): = −2.09, < .05, one-tailed, SCAS Generalised Anxiety subscale: = −2.07, < .05, one-tailed, SES Total (Self Esteem) = −2.19, < .05, one-tailed and SQC Total (Schemas) = −2.85, < .01, one-tailed.

Table 2.  Pre and post scores for participants in study 2 (= 12)
MeasurePre Mean (SD)Post Mean (SD)Z score
  1. *< .05; **< .01

 Total33.9 (11.3)30.6 (14.3)−1.33
 Separation anxiety4.7 (3.6)3.8 (2.9)−1.21
 Social anxiety5.7 (3.1)5.4 (3.6)−0.11
 Obsessive compulsive6.2 (2.2)6.5 (2.3)−0.31
 Panic/agoraphobia6.5 (3.5)6.2 (5.1)−0.40
 Physical injury fears4.2 (2.5)3.6 (2.5)−1.28
 Generalised anxiety6.8 (2.4)5.1 (2.5)−2.07*
 Self esteem (RSE)18.4 (3.5)20.1 (3.4)−2.19*
 Depression (AWS)11.2 (4.5)7.4 (5.3)−2.09*
 Schemas (SQC)65.8 (12.1)51.9 (9.4)−2.85**

Participant satisfaction

Table 3 demonstrates the median scores for the individual items on the feedback questionnaire. Average satisfaction was moderate to high, with most participants rating for example that they enjoyed TFD, it helped them with their problems, and they would recommend it to a friend.

Table 3.  Median ratings for feedback questionnaire (= 11)
ItemMedian rating
Did you enjoy TFD?
(1 not at all - 9 a lot)
Did TFD help you to understand your problems?
(1 not at all - 9 a lot)
Has TFD helped you find ways to cope with your problems?
(1 no - 9 a lot)
Have you used these new ways to cope with your problems?
(1 never - 9 all the time)
Would you recommend TFD to a friend?
(1 definitely no - 9 definitely yes)
How (difficult) did you find TFD?
(1 Too difficult, 5 just right, 9 too easy)
Was it helpful having someone with you as you used TFD?
(1 definitely no - 9 definitely yes)


These studies provide proof of the concept that cCBT (‘Think, Feel, Do’) can be successfully delivered in schools as either a universal or targeted intervention for emotional problems in children and adolescents. These initial results are promising and suggest that cCBT delivered in this way might have a positive effect upon symptoms of anxiety and low mood. However, the sample sizes in these studies are small, recruitment was opportunistic, participation flow was not recorded, and no follow-up assessments were included. Therefore, whilst encouraging, it would be premature to speculate about the possible effectiveness of ‘Think, Feel, Do’ as a school based emotional health intervention.

As previously highlighted, research into the use of cCBT with children is limited and this is the first time that the potential of cCBT as a universal emotional health intervention for children as young as 10 years has been investigated. Similarly, the use of cCBT as a structured school based intervention that requires limited training and can be delivered by non-mental health specialists offers the potential to make CBT skills more widely available. The outcomes, whilst they should be interpreted with appropriate caution, are encouraging, with the results from the targeted study being consistent with previous research in showing that cCBT can reduce symptoms of depression and generalised anxiety in high risk adolescents (Gerrits et al., 2007; Van Voorhees et al., 2008).

The qualitative comments from participants in Study 1, and the quantitative responses from Study 2 suggest moderate to high satisfaction with the cCBT intervention. Previous work has demonstrated that children and adolescents are satisfied with cCBT, often as much as face-to-face interventions (Cunningham et al., 2009; March et al., 2008). Surveys of parents and mental health professionals have also demonstrated that they believe that cCBT would be a particularly engaging intervention for young people (Stallard, Richardson, & Velleman, 2010; Stallard, Velleman, & Richardson, 2010). The participants in both studies were predominantly male. Research has previously shown that male adolescents perceive more barriers to seeking help than their female peers (Raviv et al., 2009), thus cCBT may prove especially useful with male children and adolescents who are reluctant to seek help.

In presenting these findings a number of limitations in addition to those already specified need to be acknowledged. First, the sample is predominantly male and further work is required to substantiate these findings with a larger group and to examine the potential of cCBT with girls. Second, in Study 1 there were differences in both child-rated total anxiety and parent-rated SDQ scores between the cCBT and gaming groups at baseline. Although these were not found to be significant, the findings should be interpreted with caution. Third, in both studies the researcher was not blind to the children’s condition and in Study 1 the researcher also led the sessions and feedback group, thereby raising the possibility of response bias. Finally, the absence of an appropriate comparison group in our targeted study limits the strength of any conclusions as it is unclear whether the reported improvements were due to cCBT or the passage of time.

In conclusion, these two small exploratory studies provide proof of the concept that cCBT can be delivered as an universal and targeted school based emotional health intervention. Computer-based CBT has the potential to be a widely accessible, standardised and attractive medium to deliver interventions in schools. This work also demonstrates the potential for cCBT to be delivered by non CBT specialists, a benefit that has been highlighted in work with adults (Department of Health, 2007). The level of specialist training required to facilitate cCBT consisted of one-half day workshop and therefore offers the potential to significantly increase the availability of specialist interventions. The findings of this study are promising, but limited, and should be interpreted with caution considering the limited sample sizes and restricted follow-up. Future research is warranted and should extend this work with larger samples and longer term follow-up periods.


This research was funded by the National Institute for Health Research (NIHR) under its Research for Patient Benefit programme. P.S. holds intellectual property rights for the programme ‘Think, Feel, Do’ being studied in this paper. M.A., S.M. and T.R. declare no conflicts of interest.

We would like to acknowledge the help and support provided by the head teachers of the schools that participated in the study. We would also like to thank the children who took part for their enthusiastic participation in the project and their helpful comments about the programme and its delivery. Thank you also for the work of the school nurses who delivered the intervention.

Declaration of interests

PS holds intellectual property rights for the programme ‘Think, Feel, Do’ being studied in this paper. MA, SM and TR declare no conflicts of interest.