The protocol for this RCT has previously been described in detail [22]. The study was an open two-parallel group RCT, stratified by sex, that compared a telephone-assisted ICBT initiative to an education control.

The study population consisted of children aged 10–13 years who were in grades 4–6 at comprehensive schools in the Finnish cities of Turku, Tampere and Orivesi and the counties of North Karelia and Central Ostrobothnia. We also included children in grades 5–6 in the city of Espoo. The screening started in August 2017 in Turku and was extended to the other areas in 2018, and the last participants were randomized and completed the program during the spring 2020. The children were screened during their yearly school healthcare check-ups. If they screened positive for anxiety, we contacted their families to assess their eligibility for the RCT.

The inclusion criteria included scoring at least 22 points on the 41-item child-reported Screen for Child Anxiety Related Emotional Disorders (SCARED-C) questionnaire during the school health appointment. The exclusion criteria were assessed by the research team when we contacted the parents. These included no Internet access at home, insufficient Finnish or Swedish language skills to take part and visual or hearing impairments that hindered the use of the program. Children with an intellectual disability or autism spectrum disorder were excluded from the study. We also excluded children with suicidal intentions or a severe mental health disorder, those receiving ongoing psychotherapy and those whose medication for anxiety had changed during the last two months. As a result, 234 children were randomized to the intervention group, and 233 to the education control group. The analyses described in this article only focus on the intervention group. There was one participating parent for each child and this could be either their mother or father, or, in some cases, a step parent.

The mean age of the participating children was 11.5 years and 71.1% were girls. Of the 234 participating parents, 7.7% were 34 years of age or younger, 59.0% were 35–44 years of age and 33.3% were 45 years of age or older. Just under two thirds (63.7%) of the parents had a college or university degree, while 36.3% had a lower level of education. The prevalences of each anxiety disorder in the sample are shown in Supplementary Table 1.

The school-based assessment for anxiety symptoms used five of the 41 items from the SCARED-C report [22]. Each item was scored from 0 to 2, with larger scores indicating higher levels of anxiety symptoms. The families were invited to participate in the study if the child had a total score of three or more points or scored the maximum of two points for any of the five items.

Before randomization, the parents answered questions about the family demographics, including the gender and age of their child, their own age, occupation and education level and the family’s structure and the languages the family spoke.

The parents provided details of any psychiatric symptoms that their child had by completing the Strengths and Difficulties Questionnaire (SDQ) [23]. The SDQ consists of 25 items, divided into five subscales of five questions, covering both positive and negative behaviors. The subscales are emotional symptoms, conduct problems, hyperactivity/inattention, peer relationship problems and prosocial behavior. Each item is rated on a scale of 0–2, with three possible answers: never, somewhat true and certainly true, respectively. The emotional symptoms and conduct problems subscales were considered most relevant for this study and were analyzed separately. When the parents had answered the SDQ element of the questionnaire they were asked how long the child’s difficulties had lasted and how severe the difficulties were. The SDQ has been validated in Finland and is considered reliable [24].

The full SCARED questionnaires were completed by the children (SCARED-C) and their participating parent (SCARED-P) once they had agreed to take part in the study. Both versions contain 41 items, consisting of five subscales. Four of these screen for specific anxiety disorders: panic disorder, generalized anxiety disorder, separation anxiety disorder and social anxiety disorder. There is also a subscale for school phobia, which is not classified as an anxiety disorder in the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition. The items are scored from 0 to 2, for not or hardly ever true, somewhat or sometimes true and very or often true. The maximum total score is 82. The SCARED questionnaires have been assessed in population-based samples in several countries. Four of the five subscales have proven highly reliable in cross-cultural settings, when measured by Cronbach’s alpha coefficient (alpha = 0.72–0.84). The exception is the school phobia subscale (alpha = 0.62) [25]. We have previously reported that the correlation between the child and parent reports in the Finnish SCARED is low, with children reporting higher anxiety levels than their parents [26].

Data on anxiety disorders were collected by asking the children and their parents to complete the Development and Well-Being Assessment (DAWBA) during a telephone interview. This includes both structured and open-ended questions and the answers are fed into a computer program that generates a summary sheet and predicts the most likely diagnoses. These results are then evaluated by a clinician, who confirms if the diagnostic criteria have been fulfilled [27]. The use of the DAWBA has been shown to increase the likelihood of diagnosing emotional disorders [28]. The content of the program was mainly designed to treat social anxiety and generalized anxiety disorders. As a result, we analyzed these diagnostic groups separately, in addition to the presence of any anxiety disorder in general.

The 21-item Depression Anxiety and Stress Scale (DASS-21) [29] was used to evaluate the parent’s personal stress, anxiety and depression. The items rate how well they apply to an individual, by using a four-point scale, ranging from zero for never to four for almost always. Cronbach’s alpha for the DASS-21 total has been quantified as alpha = 0.90, and the DASS-21 distinguishes well between depression, stress and anxiety [29].

After the children and their parents had completed the third iCBT module they filled in the 12-item Working Alliance Inventory - Short Revised (WAIS-SR) developed by Munder et al. [30]. When they had completed the whole iCBT program, they both filled in the eight-item Client Satisfaction Questionnaire (CSQ) from Attkisson et al. [31] online and also answered some questions about the general usability of the program at the same time.

The effect of the COVID-19 pandemic on the attrition rate was evaluated by using the date of the randomization as an indicator when the family entered the program. The 54 participants who had been randomized on 1 October 2019 or later were classified as potentially affected by COVID-19, as they completed at least a part of the program during the pandemic. Of these, 14 participants had been randomized on 1 February 2020 or later and had completed the whole program during the health crisis.

The treatment comprised nine Internet-based modules, which included digital material for both the children and their parents, and weekly calls from their coach. The web application used Django, which is an open-source web framework written in Python, and all data gathered via the electronic platform was stored in PostgreSQL database.

During the calls, the coach spoke with the parent and then spoke with the child, when the parent was present. The most crucial parts of the Internet program included text, pictures, educational audio clips and animations. Each of the nine modules focused on specific themes (Table 1), including psychoeducation (theme 1), anxiety management skills (themes 2–5), gradual exposure (themes 6–8) and the maintenance plan (theme 9). The families also received a booster call approximately one month after they finished the treatment program.

Logistic regression analysis was used to explore the risk of the families dropping out. The associations between completers and dropouts and family factors (family structure, parental age, and parental education level) as well as psychological measures (SCARED-C, SCARED-P, SDQ, DAWBA, parental DASS-21 and the child and parent versions of the WAI-SR) were performed in separate models. The statistical analyses were carried out using SAS statistical software, version 9.4 (SAS Institute Inc, Cary, North Carolina, USA). The data are reported using odds rations (OR) and 95% confidence intervals (CI).

The p-values are not corrected for the multiplicity given the exploratory nature of the study.

This study was performed in line with the principles of the Declaration of Helsinki. The study has been registered in ClinicalTrials. gov and approved by the research ethics board of the Hospital District of South West Finland (ETMK:67/1801/2017, approved on 20 June 2017). The children provided written, informed assent during the healthcare check-up and received verbal information on the study from the school nurse. The parents provided informed consent via an internet-based application that school personnel use to communicate with parents.