Cognitive behaviour therapy for pandemic-related anxiety and depression in adolescents: a pilot study of a novel intervention

ABSTRACT

Objective

Although studies have examined the efficacy of cognitive behaviour therapy for adolescents experiencing depression and anxiety during the COVID-19 pandemic, research has yet to evaluate a theory-informed intervention for pandemic-related psychopathology in this population.

Methods

A single-group pre-/post-test study design was used to examine a novel therapy for adolescents reporting pandemic-related psychopathology. To meet inclusion criteria, participants had to experience psychological symptoms attributable to, or exacerbated by, the COVID-19 pandemic, as determined by self-report and clinician judgement. A sample of 15 adolescents (M age 16.07, SD = 1.75, range 13–18; 86% female) commenced six sessions of psychological therapy. The primary outcome was pre- to post-change in psychiatric symptoms, with secondary outcomes including pandemic-related fears, behavioural withdrawal, and intolerance of uncertainty.

Results

Participants reported a significant decrease in psychiatric symptoms from pre- to post-treatment (Hedges’ g = .82) in an intention-to-treat analysis, with 60% of participants demonstrating reliable improvement by the end of treatment. Pandemic-related fears also decreased (Hedges’ g = .72), but other secondary outcomes remained unchanged at the end of the intervention.

Conclusions

This single-group study provides preliminary effect size estimates for pandemic-adapted cognitive behaviour therapy for adolescents. Further research is needed to rigorously evaluate the efficacy of pandemic-adapted CBT in this population, utilising larger sample sizes and controlled study designs.

Key points

What is already known about the topic:

(1)  Adolescents faced heightened depression and anxiety during the COVID-19 pandemic.

(2)  Standard cognitive behaviour therapy has been researched for adolescents affected by the COVID-19 pandemic, but studies on pandemic-adapted CBT for this group are notably absent.

What this topic adds:

(1)  Our study is among the first to evaluate a theory-based intervention for pandemic-related psychopathology in adolescents.

(2)  Unlike previous studies, our research focused on participants believed to be psychologically impacted by the COVID-19 pandemic, rather than those simply exhibiting symptoms aligned with the pandemic’s timeline.

KEYWORDS:

• COVID-19
• telemedicine
• cognitive behaviour therapy
• adolescents
• anxiety disorders
• depression

Global estimates of child and adolescent mental health indicate that depression and anxiety doubled during the first year of the COVID-19 pandemic (Racine et al., 2021). Although adolescents rarely develop serious illness following COVID-19 infection (Cortis, 2020), public health measures, such as school closures, had a significant negative impact on families and young people during the COVID-19 crisis. At the peak of the pandemic, adolescents reported increased levels of social isolation (Loades et al., 2020), family stress (S. H. Li et al., 2021), and contamination fears (Seçer & Ulaş, 2020). While many adolescents remained resilient in the face of these stressors (Kuhlman et al., 2021; Waite et al., 2021) at-risk adolescents experienced an escalation in psychiatric symptoms (Corrigan et al., 2022). Emerging cognitive behavioural models have identified a number of risk factors that explain the onset and maintenance of pandemic-related psychopathology (Q. Li et al., 2021; Shafran et al., 2021; Taylor, 2019). The identification of modifiable risk factors gives rise to the possibility that psychological therapies could be tailored to support adolescent mental health during epidemics and pandemics.

At the start of the COVID-19 outbreak, there were no specific psychological models of pandemic-related psychopathology. Nevertheless, a number of psychological principles had been established by 2019 based on research conducted during the 2003 SARS outbreak (Peng et al., 2010), 2009 H1N1 “swine-flu” pandemic (Brand et al., 2013; Wheaton et al., 2012), and earlier research conducted during the HIV/AIDS pandemic (Barrett et al., 1995). Converging evidence from these outbreaks suggests that personality factors play a role in pandemic-related distress, with obsessive beliefs (Brand et al., 2013) and intolerance of uncertainty (Taha et al., 2014) as key predictors of psychopathology. Furthermore, research in the 1990s and 2000s demonstrated the importance of coping style as a resilience factor during the HIV pandemic (Pakenham & Rinaldis, 2001). Subsequent research has confirmed these risk and resilience factors as predictors of COVID-19 related psychopathology in adults (Bavolar et al., 2021; Kirby et al., 2021; Paluszek et al., 2021). Studies of children and adolescents had largely been neglected prior to 2019; however, one study found that parental responses had a significant impact on the distress experienced by children during the 2009 swine flu outbreak (Remmerswaal & Muris, 2011). Research on children and adolescents has seen significant advancement during the COVID-19 period, with studies demonstrating the importance of personality (Q. Li et al., 2021), coping style (Shi & Wang, 2021), environmental reward (Wieman et al., 2022), and family factors (Wang et al., 2022) in understanding young people’s emotional difficulties during pandemics.

During the early stages of the pandemic, clinicians started to caution that the emerging outbreak might impact the mental health of vulnerable populations (Bartone et al., 2020), with several research studies developed to evaluate mental health responses to the crisis. Although there are several high-quality studies evaluating adult-focused interventions (Bonardi et al., 2022; Egan et al., 2021), fewer studies were deployed to examine adolescent interventions (Bonardi et al., 2022). In one of the few studies focusing on adolescents, Schleider and colleagues looked at the benefits of offering adolescents a single-session self-help intervention for depression during the COVID-19 pandemic (Schleider et al., 2021). Participants were randomised to either a growth mindset intervention, behavioural activation, or a nonactive control intervention. Both active interventions demonstrated significant reductions in depressive symptoms relative to a control group (d = 0.18).

Despite the positive findings, Schleider et al.’s (2021) study has a number of limitations. These limitations are highlighted below as they are common to other COVID-19 mental health studies (e.g., Chen, 2020; Duan et al., 2022; Tymofiyeva et al., 2022). Firstly, Schleider et al’.s interventions were self-help in nature – the intervention of choice in most COVID-19 studies (Bonardi et al., 2022; Egan et al., 2021). Although self-help interventions may be offered to clients in the context of stepped care (Bennett et al., 2019), therapist-delivered therapies are more common in routine practice where clinicians encounter high levels of complexity and severity (Rice et al., 2022). Secondly, Schleider et al. failed to exclude cases of depression unrelated to the pandemic or establish criteria for identifying pandemic-related psychopathology, such as pandemic-related worries, contamination fears, or lockdown fatigue. This limitation is not exclusive to their study; rather, it is similarly observed in other COVID-19 mental health studies involving adolescents (e.g., Boldt et al., 2021; Chen, 2020; Duan et al., 2022; Tymofiyeva et al., 2022), which only required participants to report experiencing anxiety and depression during the outbreak for study inclusion, without making an attempt to determine if psychological symptoms were associated with the pandemic. Finally, Schleider et al’.s treatment models were informed by conventional cognitive behavioural principles. Psychological theories that account for pandemic-related psychopathology were not integrated into Schleider et al’.s interventions. For instance, their behaviour activation protocol (Schleider et al., 2019) predated the COVID-19 pandemic, meaning it did not include adaptations to help young people cope with the crisis, such as how school closures might impact mood and valued activities. Although theoretical papers have outlined CBT adaptions for youth living through COVID-19 (Rodriguez-Quintana et al., 2021), published papers to date have only outlined protocols to test traditional treatment models such as mindfulness (Tymofiyeva et al., 2022), acceptance and commitment therapy (Duan et al., 2022), and solution focused therapy (Chen, 2020) rather than theory-informed treatment models designed specifically for pandemic-related psychopathology.

This study aims to explore psychometric changes before and after intervention on a range of measures, estimating effect sizes for future researchers interested in evaluating therapy for pandemic-related anxiety and depression in adolescents. The treatment model under investigation was informed by psychological principles that help to explain emotional difficulties experienced during pandemics (Taylor, 2019). Inclusion criteria required participants and intake clinicians to attribute psychological symptoms to the pandemic, with the primary outcome being changes in psychiatric symptoms. Secondary outcomes included behavioural withdrawal, COVID-19 fears, and intolerance of uncertainty, as these constructs informed the treatment model theoretically. The cognitive-behavioural intervention was delivered by therapists via telehealth, with a protocol incorporating COVID-19 adaptations specifically tailored to support adolescent mental health during a pandemic. This pre-registered study refrained from making specific predictions, apart from anticipating symptom reduction, due to the exploratory nature of the research.

Method

Design

The study was a single group pre-/post-test design. Ethical approval was received from the Swinburne University of Technology Ethics Committee (Approval number 20202937–4543) and the study’s method, aims, and hypotheses were preregistered through the Open Science Framework (https://osf.io/aqb5w/).

Recruitment and procedure

Recruitment commenced in June 2020 with schools and health services in Melbourne and the state of Victoria. The program was run at the Swinburne University Psychology Clinic. The program was offered to the community as a low-fee therapy during the COVID-19 pandemic. Referrals were assessed by an intern psychologist under the supervision of a senior clinical psychologist (principal author). Assessment involved a clinical history, psychometric measures, and a semi-structured diagnostic interview. Eligibility for the program was determined by the treating clinician in consultation with the principal researcher after completion of the clinical assessment. When participants were informed that they met eligibility criteria for therapy, a student researcher contacted participants to see if they would like to also participate in a study evaluating the program’s effectiveness. Although participants could receive the clinical intervention without participating in the evaluation, all eligible adolescents consented to participate in the study. Parental consent was also obtained for all adolescents 17 years and under (n = 11). The protocol entailed weekly sessions during the initial 5-week period, followed by a fortnight between the 5th and 6th sessions. However, strict adherence to this protocol was not always feasible due to participants occasionally rescheduling appointments or having other commitments.

Participants

Participants were 15 adolescents aged 13–18 presenting to a university psychology clinic during the COVID-19 pandemic with anxiety and depression. To be included in the study, participants’ mental health difficulties had to be caused or exacerbated by the pandemic. This was determined by (a) clinician judgement during the initial assessment and (b) participant self-report on a 4-point likert scale (i.e., “How much has the Coronavirus situation had an impact on your mental health?”, 1 = “Not at all” and 4 = “A great deal”). Although it is not technically possible to definitively establish whether the pandemic “caused” or “exacerbated” participants’ mental health difficulties, these judgements were made by the assessment clinicians using clinical history taking and relevant clinical data, under the careful supervision of the primary author. Other inclusion criteria included (a) elevations on one or more subscales on the Revised Children’s Anxiety and Depression Scale (RCADS); (b) aged 13–18; and (c) parental consent to participate in the program. Exclusion criteria encompassed the following: (a) moderate-to-severe mental health difficulties; (b) PTSD; (c) psychosis; and (d) elevated suicide or self-harm risk. Subclinical and excluded cases were identified through clinical judgement, supplemented by the MINI-KID assessment, clinical interview, and psychometric data. The intentional inclusion of subclinical cases in the study aimed to represent the broad spectrum of adolescents seeking mental health support during the COVID-19 crisis. All participants accessed the program between 17 July 2020 to 18 March 2021. This period coincided with a series of COVID-19 outbreaks in Melbourne Australia, a city that implemented strong lockdowns and restrictions to achieve “COVID zero” (Collyer et al., 2022; Smith, 2020). During the 244-day study period, approximately 107 days occurred during lockdown, during which movement outside of one’s home was restricted to essential activities such as medical care, grocery shopping, and limited exercise (Dunstan, 2021). Additionally, there were approximately 10–16 weeks of onsite schooling during the 8-month study period.

Intervention

The cognitive behavioural intervention was developed by the lead author in April 2020 as a response to the anticipated impacts of the COVID-19 pandemic on adolescent mental health. The intervention was theoretically informed by psychological research conducted during past epidemics and pandemics (Taylor, 2019). The broader psychological literature was also considered in the development of the intervention, including cognitive behavioural treatment principles, telehealth service models, and adolescent resiliency research. Following this review, it was reasoned that pandemic-related depression, anxiety, and coping among adolescents might be best understood through the application of the following psychological principles:

1. Lockdowns expose adolescents to low reward environments – a risk factor for depression (Carvalho et al., 2011; Pass et al., 2018; Wieman et al., 2022);

2. Intolerance of uncertainty and obsessive beliefs are psychological risk factors for pandemic-related anxiety (Brand et al., 2013; Q. Li et al., 2021; Taylor, 2019);

3. Effective problem solving is a protective factor for adolescent mental health during a crisis (Frye & Goodman, 2000; Shi & Wang, 2021; Spence et al., 2002); and

4. Parental responses may exacerbate a young person’s pandemic-related worries (Remmerswaal & Muris, 2011).

Accordingly, we developed a six-session therapy program, delivered via video-based telehealth, to specifically address low reward environments, pandemic appraisals, pandemic-related stressors, and family support. Cognitive behavioural strategies included behaviour activation (Pass et al., 2018), cognitive restructuring (Clark, 2013), and problem solving skills (Kennard et al., 2009), implemented within the framework of behavioural family treatment (Reuman et al., 2021; Wells & Albano, 2005). To illustrate, the behaviour activation session explores the relationship between low reward environments and depression, the effects of lockdowns on mood, and how young people can take steps to structure their days during lockdown with activities that are likely to be rewarding. Psychometric feedback was integrated into the intervention, with therapists utilising questionnaire data to inform formulation, track progress, and raise client awareness (Tam & Ronan, 2017). The Goal-Based Outcomes tool (Law & Jacob, 2015) and a child version of the Negative Problem Orientation scale (Perrin et al., 2019) were employed for clinical purposes, specifically for treatment planning and client feedback. However, it is important to note that they were not part of the pre-/post-test study analysis. Clinicians (N = 9) working in the program were provisionally registered psychologists with the Psychology Board of Australia and received weekly group supervision from the principal author. An overview of the program’s session structure is displayed in Table 1, and worksheets and psychoeducational materials are available at the project’s Open Science Framework page.

Table 1. Program overview.

Session 1. Goal development and program philosophy
Program overview Developing program goals (i.e., GBOs) Psychometric feedback: RCADS and Fear of COVID-19 Scale
Session 2. Behaviour activation with adolescent and parent(s)
Psychoeducation regarding the connection between lockdowns, low reward environments, and mood Behaviour activation Psychometric feedback: BADS-SF
Session 3. Cognitive biases and formulation
Psychoeducation on thinking styles (e.g., intolerance of uncertainty) that elevate anxiety during pandemics Sharing CBT formulation with adolescent and parent(s) Psychometric feedback: OBQ-44 and IOU
Session 4. Challenging and testing thoughts
ABC-D model Behavioural experiments Homework task using disputation or a behavioural experiment on unhelpful pandemic-related thoughts
Session 5. Problem solving
Psychoeducation on problem solving and its protective nature during a pandemic Problem solving exercise Psychometric feedback: NPO
Session 6. Review and relapse prevention Progress review using the RCADS and GBOs Relapse prevention plan

Note. GBO = Goal Based Outcomes (Law & Jacob, 2015); RCADS = Revised Child Anxiety and Depression Scale; IoU = Intolerance of Uncertainty Scale; BADS-SF = Behavioral Activation for Depression Scale-Short Form; OBQ-44 = Obsessive Beliefs Questionnaire; NPO = Negative Problem Orientation Scale – child version (Perrin et al., 2019).

Measures

The following measures were used to assess the baseline characteristics of the sample (Note: reported internal consistency values are from the current study’s sample):

Categorical diagnosis was determined using four modules (i.e., depression, generalised anxiety, separation anxiety, and panic disorder) from the Mini International Neuropsychiatric Interview for Children and Adolescents (MINI-KID) (Sheehan et al., 2010). The MINI-KID has comparable reliability and validity when compared with other diagnostic interviewing tools (Petts et al., 2016). MINI-KID scoring rules were followed for reaching diagnostic conclusions, but clinician judgement was also used so that the MINI-KID supplemented clinical assessment rather than replacing it. Therapists also had the option to record a “subclinical” diagnosis if clients presented with a significant number of symptoms but failed to reach MINI-KID thresholds for a categorical diagnosis.

Cognitive risk factors hypothesised to place adolescents at risk of psychopathology during a pandemic were measured with the Obsessive Beliefs Questionnaire – Child Version (OBQ-CV) The 44-item measure has good internal consistency, retest reliability, and convergent validity The OBQ-CV has three subscales: responsibility/threat (16 items; α = .94), perfectionism/certainty (16 items; α = .89), and control (12 items; α = .91). For descriptive purposes, subscales were considered “elevated” when scores were ≥ 1 SD above the mean of a comparison sample (Coles et al., 2010).

The following measures were used to assess pre-to-post intervention change:

Psychiatric symptoms were assessed with the Revised Child Anxiety and Depression Scale (RCADS) (Chorpita et al., 2000). The RCADS is a 47 item self-report questionnaire for children and adolescents aged 7–18 years with good psychometric properties (Chorpita et al., 2005; de Ross et al., 2002). The RCADS total score (47 items; α = .97) represents general psychopathology, with a range from 0–141. The RCADS total score is comprised of six underlying subscales: depression (10 items; α = .86), generalised anxiety (6 items; α = .90), obsessive-compulsive symptoms (6 items; α = .79), panic (9 items; α = .86), separation anxiety (7 items; α = .89), and social phobia (9 items; α = .84). The scores on the RCADS were considered “elevated” if they were ≥ 1 SD above the mean of the Chorpita et al. (2000) validation sample.

Pandemic-related worries were assessed with the Fear of COVID-19 Scale (FCV-19S) (Ahorsu et al., 2020). The FCV-19S measures fears, worries, and anxieties related to COVID-19. The scale has good psychometric properties (Ahorsu et al., 2020), with high internal consistency in the current sample at baseline (7 items; α = .92). Scores on the scale range between 7–35, with participants rating their level of agreement on each item from 1 (strongly disagree) through to 5 (strongly agree). Example questions include “I am afraid of losing my life because of coronavirus-19”, and “I cannot sleep because I’m worrying about getting coronavirus-19”. Additionally, for descriptive purposes, scores were classed as “elevated” when ≥ 1 SD above the mean of an adolescent comparison sample (Seçer & Ulaş, 2020).

Behavioral activity levels were assessed with the Behavioral Activation for Depression Scale-Short Form (BADS-SF) The BADS-SF is a 9-item scale with responses rated from 0 (not at all) to 6 (completely), with higher scores indicating greater levels of activity. Example items include, this week “I engaged in many different activities” and this week “I did things that were enjoyable”. Scores on the scale range from 0–54, with good internal consistency in the current sample at baseline (9 items; α = .83). The BADS-SF has demonstrated adequate psychometric properties in adolescent samples Additionally, scores were classed as “elevated” when ≥ 1 SD above the mean of a comparison sample (Petts et al., 2016).

Intolerance of uncertainty was assessed using the brief Intolerance of Uncertainty (IoU) scale The 5-item IoU is an adaptation of the 27-item scale for adults (Buhr & Dugas, 2002), with the 5-item version being designed to be developmentally more appropriate for children and adolescents Example items include, “I can’t be relaxed if I don’t know what will happen tomorrow” and “Not knowing what may happen next makes my life horrible”. The scale has good psychometric properties, with the 5-item and the 27-item scales appearing to measure a similar underlying construct In the current sample, the IoU scale displayed good internal consistency at baseline (5 items; α = .88), Additionally, IoU scores were classed as “elevated” when ≥ 1 SD above the mean of an adolescent comparison sample (Fialko et al., 2012).

Analyses

Descriptive statistics for each measure and demographic variable at baseline are provided. This includes the percentage of participants scoring in an “elevated range” at baseline on each clinical measure. An intent-to-treat analysis (n = 15) was conducted with the last observation carried forward (LOCF). Although it has been highlighted that LOCF has some disadvantages compared with statistical imputation (Salim et al., 2008), a sensitivity analysis on the current sample indicated that LOCF was more conservative in estimating treatment effects when compared with expectation-maximisation (iterations = 25) (Schlomer et al., 2010). Additionally, a complete case analysis (n = 12) resulted in higher effect size estimates (Hedges’ g) for all outcomes (RCADS = .96; FCV-19S = .83; IoU = .33; BADS-SF = −.48), in comparison to the LOCF method. While a variety of options exist for handling missing data (Schlomer et al., 2010), this study opted for a conservative approach to avoid overestimating clinical effectiveness. Specifically, dropouts were treated as non-responders using LOCF to ensure a cautious evaluation of the results, an approach taken by other pilots using pre-/post-test study designs (e.g., Curran et al., 2021; Pass et al., 2018). Mean difference scores were tested using paired samples t-tests and converted into corrected standardised effect sizes (Hedges’ g; Durlak, 2009). Mean difference scores that violated the assumptions of normality were followed up with robust tests (Field & Wilcox, 2017) using Yuen’s test on trimmed means for dependent samples, which was based on 20% trimming. A three-month follow-up was not analysed because completion of the online questionnaire was too low to be statistically meaningful (n = 5); however, the recovery status of each case in the follow-up sample is reported. To examine clinically meaningful change, a Reliable Change Index (RCI) was calculated for the pre-to-post-test intervention scores (see Table 3), which indicates change greater than measurement error (Jacobson & Truax, 1991). Descriptive statistics, effect size calculations, and parametric tests were performed on SPSS statistical software (V.27, IBM). Robust tests, normality assumptions, psychometric properties, and reliable change calculations were performed using R software, V.4.0.3 (R Project for Statistical Computing).

Table 2. Baseline sample characteristics (N = 15).

Variable	n (%)	Sample Range	Mean	SD
Age		13–18	16.07	1.75
School Type
State	13 (86.67)
Private	2 (13.33)
Gender
Male	2 (13.33)
Female	13 (86.67)
Aboriginal/Torres Strait Islander status	0 (0.00)
Family conflict	5 (33.3)
Psychiatric medication (current)	6 (40.00)
No previous MH treatment	4 (26.77)
MINI KID Diagnosis
Depression	7 (46.67)
GAD	1 (6.67)
Separation Anxiety	1 (6.67)
Panic Disorder	1 (6.67)
Subclinical anxiety & depression	5 (33.33)
Elevated RCADS Scores^a	8 (53.33)
Depression	10 (66.67)
GAD	6 (40.00)
OCD	4 (26.67)
Panic	7 (46.67)
SAD	6 (40.00)
SOC	5 (33.33)
Elevated OBQ Scores^a
Responsibility/Threat	5 (33.33)
Perfectionism/Certainty	6 (40.00)
Control	4 (26.67)
Elevated Fear of COVID Scores^a	7 (46.67)
Elevated IoU Scores^a	9 (60.00)
Elevated BADS-SF Scores^a	5 (33.33)

Note. MINI KID = Mini-International Neuropsychiatric Interview for Children and Adolescents; RCADS = Revised Child Anxiety and Depression Scale; GAD = Generalized Anxiety Disorder; OCD = Obsessive Compulsive Disorder; SAD = Separation Anxiety Disorder; SOC = Social Anxiety; IoU = Intolerance of Uncertainty; BADS-SF = Behavioral Activation for Depression Scale-Short Form.; OBQ = Obsessive Beliefs Questionnaire; IoU = Intolerance of Uncertainty Scale.

^aScores classified as “elevated” when ≥ 1 SDs above norm-referenced sample (see measures section for reference group.

Table 3. Intention-to-treat symptom outcomes (n = 15).

Measure	Baseline M (SD)	Posttreatment M (SD)	Change from baseline (95% CI)	t	p	Effect size (Hedges’ g)	RCCrit	Reliable Improvement (%)
RCADS Total	57.67 (28.63)	39.53 (21.36)	18.13 (6.17–30.10)	3.25	.006	.82	14.85	60.00
Depression	15.13 (6.38)	10.67 (4.55)	4.50 (1.05–7.90)	2.81	.014	.71	6.69	26.67
GAD	8.87 (4.85)	5.60 (4.08)	3.27 (0.72–5.81)	2.76	.015	.69	4.36	40.00
OCD	5.07 (3.22)	3.13 (2.72)	1.93 (0.10–3.76)	2.27	.040	.57	4.09	20.00
Panic	8.73 (6.40)	6.33 (5.01)	2.40 (0.07–4.73)	2.21	.044	.56	6.64	20.00
SAD	6.20 (5.75)	4.07 (4.54)	2.13 (0.12–4.15)	2.27	.040	.57	5.29	13.33
SOC	13.67 (6.17)	9.73 (5.36)	3.93 (1.27–6.60)	3.16	.007	.79	6.75	20.00
FCV-19S	18.13 (7.71)	13.20 (5.85)	4.93 (1.24–8.63)	2.86	.012	.72	6.19	33.33
IoU	16.93 (4.99)	15.07 (5.9)	1.87 (−1.34–5.07)	1.25	.232	.31	4.79	20.00
BADS-SF	24.27 (12.06)	30.33 (12.0)	−6.07 (−12.54–1.41)	−1.74	.104	−.44	−13.7	13.33

Note. All psychometric scores are raw scores; RCCrit = Reliable Change Criterion; RCADS = Revised Child Anxiety and Depression Scale; GAD = Generalized Anxiety Disorder; OCD = Obsessive Compulsive Disorder; SAD = Separation Anxiety Disorder; SOC = Social Anxiety; FCV-19S = Fear of COVID-19 Scale; IoU = Intolerance of Uncertainty; BADS-SF = Behavioral Activation for Depression Scale-Short Form.

Results

Participant flow

Figure 1 illustrates the participant flow during the study. Twenty-four participants enquired about the CBT-PDAS program at the university clinic. Most of these enquiries came from parents as parental consent was required for those under the age of 18. Twenty-one young people were invited to a clinical assessment and a semi-structured interview (MINI KID) over telehealth by an intern psychologist, with two participants failing to engage with the assessment process. At the end of the assessment, two failed to meet inclusion criteria and two others preferred to seek treatment from other services. Fifteen participants were deemed appropriate for the intervention, and all agreed to commence and complete the baseline measures. During the course of the intervention, three participants dropped out, meaning 12 participants completed posttreatment measures. Posttreatment measures were completed as a homework task between sessions five and six, in order to obtain scores that could be used during the final session to develop a relapse prevention plan. On completion of the program, participants were sent an automated three-month follow-up symptom questionnaire. Compliance with the follow up questionnaire was low with only five participants completing it.

Figure 1. Participant flowchart.

Baseline characteristics

Fifteen participants started therapy and completed baseline measures (see Table 2). The sample was aged 13–18 (M = 16.07) which was composed of two males and thirteen females. The majority of the adolescents were enrolled in state-based schools (87%) and none were from an Aboriginal or Torres Strait Islander background. The most common diagnosis on the MINI KID was depression (47%) followed by subclinical anxiety and depression (33%). Fifty-three percent of the sample had elevated (≥1 SDs above norm-referenced sample) scores on a broad-spectrum anxiety and depression measure (RCADS) with the majority of the participants (67%) reporting an elevation on the depression subscale. Although only around half of the participants reported elevated COVID-19 specific fears (47%), all participants reported that their mental health had been affected “quite a lot” or “a great deal” by the COVID-19 pandemic on a likert scale during baseline assessments. Treatments were commenced and completed by participants between July 2020 – March 2021.

Outcomes

Primary outcome

Participants’ total anxiety and depression scores on the RCADS demonstrated a statistically significant mean decrease from 57.67 (28.63) pretreatment to 39.53 (21.36) post-treatment, a mean decrease of 18.13, 95% CI (6.17 to 30.10), t(14) = 3.25, p = .006, g = .82. This represents a large effect size with 60% of participants classed as reliably improved. Additionally, statistically significant improvements were observed on all RCADS subscales (p < .05) with moderate pre-to-post treatment effects (see Table 3). The SOC subscale showed departures from normality, but a robust test confirmed similar results to the parametric test, Y_t = 3.70 (1.47, 6.31), p = .006, r = 0.48. Reliable improvement outcomes were more modest on the individual RCADS subscales, ranging from 13–40%. Two participants recorded change on the RCADS total score that were classified as “unreliable”. There were also some indications of “reliable deterioration” on the RCADS subscales, but this only occurred for two participants. One participant reported deterioration on both the GAD and the OCD subscales, and another participant reported deterioration on the depression subscale.

Secondary outcomes

Table 3 also displays baseline and post-treatment scores for the secondary outcome measures: the fear of COVID-19 scale (FCV-19S), intolerance of uncertainty (IoU), and behavioural activation scale (BADS-SF). All difference scores for the secondary measures showed departures from normality, so robust tests were performed. For ease of interpretation, Table 3 only shows the test statistics and probability levels for the parametric tests, as the robust tests showed similar results. Yuen’s test on trimmed means for dependent samples based on 20% trimming indicated a statistically significant result for COVID-19 fears from baseline to posttreatment, Y_t = 4.45 (0.28, 9.50), p = .040, r = 0.46, a medium effect size. This represents a 33% reliable improvement in COVID-19 fears across the two timepoints. The intolerance of uncertainty and the behavioural activation scales were not statistically significant.

Three-month follow-up

Reliable improvement status for the primary outcome remained stable from posttreatment to follow-up (n = 5), with three participants classed as “improved” and two “unchanged” at both timepoints.

Discussion

In this study, the primary objective was to observe the extent of symptom change in adolescents who received a theory-informed intervention for pandemic-related anxiety and depression. The study aimed to determine preliminary effect size estimates, thereby contributing valuable data to support future researchers in their evaluation of psychological therapies for this clinical population. Results showed large and statistically significant changes in psychiatric symptoms (Hedges’ g = .82) and pandemic-related worries (Hedges’ g = .72) from baseline to posttreatment. The reduction in psychiatric symptoms represented a 60% “response rate” as defined by the reliable change criterion (Jacobson & Truax, 1991). Pandemic-related worries were less responsive, with only 33.3% of participants recording a reliable improvement. Although constructs associated with the treatment model (i.e., behavioural withdrawal and intolerance of uncertainty) showed a slight improvement over time, these changes were small and statistically nonsignificant. The novel therapy appeared acceptable to participants, with 80% completing the intervention. Psychiatric symptoms remained stable at follow-up; however, few participants completed the final questionnaire (n = 5).

Despite the single-group design, which precludes attributing symptom change to the intervention, this study has a number of strengths. Firstly, the study sample comprised exclusively of adolescents who experienced mental health problems arising from the COVID-19 crisis, as determined by clinician evaluation and participant self-report. Other intervention studies (e.g., Schleider et al., 2021), in contrast, failed to determine if psychopathology was related to the COVID-19 pandemic. Before COVID-19, there was a high prevalence of depression and anxiety among adolescents (Lawrence et al., 2016). A strength of this investigation’s approach lies in its deliberate effort to exclusively include only individuals experiencing pandemic-related psychopathology, despite the inherent challenge of making such judgements. A second strength of the study was the inclusion of an outcome measure to monitor pandemic-related fears during treatment. Although some studies have used pandemic-related measures to evaluate adult (Bonardi et al., 2022) and child (Guzick et al., 2022) interventions, to the best of our knowledge, no studies with adolescents have used validated COVID-19 measures. Duan et al. (2022) employed the two-item GAD-2 scale with adolescent participants, rephrasing the items to be applicable to the COVID-19 context; however, this was a custom-made scale with unexamined psychometric properties. Finally, another strength of the study was that it investigated a modified version of CBT adapted for the COVID-19 pandemic. In comparison, other studies conducted during the pandemic offered adolescents unmodified versions of CBT, such as mindfulness (Tymofiyeva et al., 2022) or behavioural activation (Schleider et al., 2021), rather than an intervention that was theoretically informed and tailored specifically to the COVID-19 situation. The current CBT intervention had a focus on parental involvement, problem-solving, managing uncertainty, and increasing environmental reward, adding to an emerging evidence base for the significance of these factors for adolescent wellbeing during pandemics (Shafran et al., 2021; Shi & Wang, 2021).

The current study has a number of limitations. Firstly, no objective measures or checklists were used to monitor therapist adherence to the treatment protocol. While a manual was used to facilitate the standardisation of treatment delivery across therapists (Brunk et al., 2015; Kendall & Frank, 2018), we lack data to confirm adherence to the manual. Secondly, many participants scored in the nonclinical range at baseline. While this did not affect tests of statistical significance or effect size calculations, it prevented the calculation of more robust metrics of clinical significance (Jacobson & Truax, 1991). Clinical assessment indicated that participants scoring in the nonclinical range at baseline experienced distress and impairment. Therefore, their inclusion in the study seemed justified. Thirdly, the follow-up questionnaire was only completed by a small number of participants, so it is unclear whether gains were maintained after treatment. Lastly, the study’s single group pre-/post-test study design and small sample size means the results should be viewed with caution.

The current study was conducted during a “once-in-a-lifetime” global pandemic. Given this context, the study has a number of important clinical implications. Firstly, the findings indicate that CBT with minor modifications is a promising approach for addressing pandemic-related psychological problems among adolescents. Key strategies included behaviour activation, behavioural experiments, tailored psychoeducation, and problem solving. Secondly, psychological theory was rapidly deployed in the current study to solve a problem of major public concern (Golberstein et al., 2020). At the start of the pandemic, evidence-based therapies were unavailable for pandemic-related psychopathology. The study’s intervention development was informed by principles from the scientist-practitioner model (Hayes et al., 1999), public health frameworks (Wight et al., 2016), and psychological research conducted during prior pandemics (Taylor, 2019). In the future, society will continue to face novel health and social problems for which evidence-based solutions are unavailable. Dissemination science is less prepared for emerging problems, as evidence-based guidelines are informed by established treatments and randomised controlled trials (McHugh & Barlow, 2010). Accordingly, practitioners should be encouraged to utilise the principles and frameworks employed in the development of the current intervention (e.g., scientist practitioner model) to address future novel challenges that will continue to arise in healthcare, social welfare, and mental health.

Overall, the study demonstrates the potential benefits of adapting cognitive behaviour therapy for adolescents living through a pandemic, with participants in the current study reporting reductions in psychiatric symptoms and pandemic-related worries. Pandemic risk is an ongoing concern for the world with continued environmental changes, the expansion of livestock production, and increases in the hunting and trading of wildlife (DiMarco et al., 2020). The COVID-19 pandemic has shown that infectious disease outbreaks have wider implications than just physical health, with far-reaching psychosocial consequences that detrimentally impact the mental health of affected communities. To ensure the provision of tailored and effective interventions to young people during future infectious disease outbreaks, it is essential to continually develop and evaluate pandemic-adapted CBT. This includes the utilisation of larger sample sizes and controlled study designs, along with integrating the most up-to-date scientific findings on the psychology of pandemics during intervention development.

Open Scholarship

This article has earned the Center for Open Science badges for Open Materials and Preregistered. The materials are openly accessible at https://osf.io/aqb5w/ and https://osf.io/2gyvz

Acknowledgements

I would like to express my gratitude to Professor Luke Downey for his generous support and mentorship throughout my research project as an Early Career Researcher at Swinburne University of Technology.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Correction Statement

This article has been corrected with minor changes. These changes do not impact the academic content of the article.