The PT trial is designed as a randomized, controlled, experimental, open, single center, with three-arm parallel groups. Randomization has been blocked with a 1:1 allocation.

The study was evaluated and approved by the local and the National Ethical Committee; each included family was informed and gave written consent agreeing to participate in the study. The study protocol was registered in the REBEC platform, which is referenced by the clinical trials (UTC number U1111-1293-9285). The study follows the CONSORT principles and statement (Schulz et al., 2015).

This clinical trial was funded by the Coordination for the Improvement of Higher Education Personnel (CAPES); the Research Program for the SUS (PPSUS: Foundation for Research Support of the State of Minas Gerais; Secretary of State for Health of Minas Gerais; Ministry of Health Brazil); and the National Council for Scientific and Technological Development. This funding source had no role in the design of this study and will not have any role during its execution, analyses, interpretation of the data, or decision to submit results. The author(s) declare(s) that they have no competing interests.

One hundred and thirty-two children were screened at the Impulsivity and Attention Research Center between March/2021 and May/2022 and were enrolled in our trial after meeting the pre-established inclusion/exclusion criteria. The collection date followed what was proposed in the research project approved by the funding agencies. Fifty-seven children and their families fully participated in the study. Families whose children were boys, aged between 6 and 12 years, and who had externalizing symptoms of hyperactivity/impulsivity and/or defiant behavior participated in the study. Exclusion criteria are as follows: ADHD inattentive without externalizing symptoms; families whose children or caregivers scored below the 5th percentile in standardized IQ tests; children with severe genetic or neurological conditions or with severe psychiatric comorbidities (i.e., psychosis, severe depression, severe autism, bipolar affective disorder); families whose caregivers had a reported severe psychiatric diagnosis (i.e., severe depression, psychosis, and bipolar affective disorder); or have few years of formal education. In addition, families with severe social adversity, such as constant exposure to hunger, violence, or extreme poverty, were also excluded from this research (Kazdin, 2005).

As part of the usual follow-up, families underwent an initial assessment consisting of a clinical interview based on the DSM-5. The interview was conducted by a psychologist from the center with experience and expertise to conduct it. The child and the primary caregiver underwent an intelligence assessment, and the child was also assessed for school performance. As patients were included in the study, permuted blocks of six participants/families were formed, which were randomized 1:1 into three parallel groups stratified by children’s age. Eleven blocks were formed, totaling 66 participants. Of these, 57 concluded their participation in this research, and nine were excluded due to decline in treatment (five; three of GROUP 03 and two of GROUP 02); diagnostic divergence between the screening team and the psychiatric team (one of GROUP 01); recently conducted parent training (one of GROUP 02); duplicate enrollment and screening (one of GROUP 01); and incomplete intervention until completion of data collection (one of GROUP 03). The 1:1 randomization assigned two participants to each of the three groups:

GROUP 01 (n = 20): Waiting list for behavioral approach while under standard treatment (ST). Bi-monthly consultations with a child and adolescent psychiatrist, including drug treatment if there is a clinical indication (at medical discretion, according to the clinical protocol), and without any other type of complementary treatment.

GROUP 02 (n = 19): Standard treatment and behavioral PT intervention in face-to-face format (ST + Face-to-Face PT). Bi-monthly consultations with a child and adolescent psychiatrist, including drug treatment if there is a clinical indication (at medical discretion, according to the clinical protocol), and complementary behavioral intervention: parent management training, in face-to-face format, with a specialized therapist, in six sessions held on a weekly basis, and adapted from the “Parent Management Training” manual developed by Kazdin (2005).

GROUP 03 (n = 18): Standard treatment and behavioral PT intervention in online format (ST + online PT). Bi-monthly consultations with a child and adolescent psychiatrist, including drug treatment if there is a clinical indication (at medical discretion, according to the clinical protocol), and complementary behavioral intervention: parent management training, in an online format, on a platform developed for the study, in six modules, to be carried out on a weekly basis, and adapted from the “Parent Management Training” manual developed by Kazdin (2005).

Concurrent behavioral interventions were not allowed. All families provided consent and assent forms.

After randomization, the participants underwent a pre-intervention assessment, filled out digitally and in person at the outpatient clinic. After that, they were immediately referred to the first consultation with the childhood and adolescent psychiatry medical team. At the beginning of the medical follow-up, if there was a divergence of diagnosis in relation to screening assessment, the team discussed the case and if necessary, the participant could be excluded, maintaining the possibility of accessing our online parent training. Soon after the pre-intervention assessment and the beginning of the medical follow-up, the allocation of the participant was revealed to the specialized therapist responsible for the interventions, who contacted the family to schedule the start of the face-to-face or online intervention. The control group waited for about 6 weeks and then was directed to the post-evaluation. Upon completion of the post-assessment, control participants had access to online parenting training. The other groups underwent the same post-assessment at the end of the interventions. For participants’ CONSORT flowchart, see Figure 1.

Parent training was planned based on Kazdin’s (2005) directions but modified for six sections. Six face-to-face sessions and six analogs online modules were developed, in which management strategies are taught to parents, with the aim of increasing caregivers’ control over children’s behavior, teaching and strengthening adaptive behaviors, and putting potentially harmful behaviors into extinction. The intervention proposed weekly appointments, but there were variations between the groups regarding the time for completion due to reschedule needs. In the online sessions/modules, the content was exposed by a short film with common examples, and at the end, support materials were made available, which always had a booklet summarizing the seen content and other materials, such as a chart of illustrations for that the caregiver can visually assemble the routine with the child.

In brief, in session/module 01, the following points are worked on: ABC of behavior; the importance of clarity in describing problem behaviors and the concept of positive opposite behavior; efficient strategies when giving commands; and routine and the importance of consistency. In session/module 02: The reinforcing consequences; the power of praise and a more enthusiastic way of delivering it that more efficiently selects behavior for having a greater magnitude; and the incentive board as a very useful and dynamic tool. In session/module 03: punitive consequences as a complement to reinforcement strategies; time out reinforcement; the strategy of ignoring bad behavior; the reprimand for behaviors that need to be stopped; and the loss of privileges. In session/module 04: modeling as a systematic teaching process, necessary for the effectiveness of interventions in many cases—adapting the level of demand is part of the proposal to create contingencies so that the appropriate behavior occurs and can be reinforced. In session/module 05: Review and punitive task for more serious and atypical behaviors. In session/module 06: Conflict resolution/negotiation + application of learned strategies to example situations.

All sessions/modules have a summary booklet, in addition to the following materials: behavior record sheet and routine illustrations kit (mod. 01); incentive board and poster for parents, showing the main strategies when giving commands (mod. 02); punishment planning sheet (mod. 03); sheet for modeling planning and modeling sheet for school (mod. 04); and cards with step-by-step and trading rules (mod. 06). The download is available upon completion of the module for all online participants and was delivered in print to face-to-face participants.

On the online platform, the intervention is 100% self-directed, with no contact with the therapist, and the videos already seen can be watched as many times as the participant needs. There is an introductory video with the main therapist explaining the intervention, and the use of the platform and an email address are available for operational questions. The content is exposed through animations, narrated by the main therapist of the project, and divided into several sequential videos of a maximum of 10 min each (the number of videos and durations may vary according to the module). Once one module is complete, the following content is only released after a week, and the day before completing a week, parents receive teasers about the content to be seen the next day.

The face-to-face intervention was carried out by psychologists from the Research Center of Impulsivity and Attention (NITIDA), previously trained and with at least 5 years of experience in behavioral psychology. For the two intervention groups, the maximum distance allowed between sessions/modules was 60 days, considering the feasibility of the project and participants being always encouraged to complete it weekly. In addition to the teasers, the technician responsible for the online platform is kept in direct contact with the online participants to remind them of deadlines and ask about technical difficulties. To ensure that the caregiver watched and absorbed the content, a quick test covering what was worked on was requested after the completion of each online module, requiring at least 60% success for participants. The duration of the sessions/modules varies, ranging from 1 h to 1 h 30 min in the face-to-face format (depending on the content, caregivers’ doubts, and dynamics of the session) and from 20 to 40 min in the online format (depending on the content).

From the pre-evaluation to the start of the intervention/waiting time and from the end of the intervention/waiting time to the post-evaluation, the maximum interval was 60 days, considering the feasibility of the project and being scheduled as soon as possible according to the availability of the patients. The post-assessment consisted of a battery of validated scales to be filled out digitally, in person at NITIDA, in addition to a new clinical interview (same conducted in screening) focusing on the symptoms of ADHD and ODD, applied by a team of specialized psychologists, blinded to the group. The entire team responsible for recruitment and evaluations was blinded to the type of intervention performed on each participant. Descriptive statistics regarding sociodemographic and clinical variables of the children and their parents are shown in Table 1.

For the screening process and semi-structured diagnostic investigation applied by a trained professional, the Kiddie Schedule for Affective Disorders and Schizophrenia for School Aged Children—Lifetime Version (K-SADS-PL; clinical interview) was used, which is based on the DSM-5 criteria (American Psychiatric Association, 2014). The interview investigated all symptoms described in the Diagnostic and Statistical Manual of Mental Disorders (DSM-5) and was applied to the primary caregiver by a trained psychologist and a member of NITIDA. For the assessment of the caregiver’s and the child’s intelligence, the Vienna (Waldherr et al., 2014) and Raven matrices tests (Pasquali et al., 2002) (respectively) were used, being both validated and standardized in the Brazilian context, and are intended to assess fluid intelligence or logical reasoning.

The pre-intervention assessment had a battery of scales and questionnaires to be filled in by the caregiver, some of which were self-reported and others related to the child’s behavior, in addition to a standardized socioeconomic status questionnaire [Associação Brasileira de Empresas de Pesquisa (ABEP), 2015]. All the procedures were done according to the previously published protocol (Chequer de Castro Paiva et al., 2022). Caregivers filled out the following scales:

The questionnaires related to the expected outcomes in this trial and reapplied in the post-intervention evaluation were MTA-SNAP-V, PSSP, SDQ, WHOQOL-BREF, and Kidscreen.

Participants were randomly assigned to either control or one of the experimental groups with a 1:1 allocation as per a computer-generated randomization schedule. The allocation sequence was generated applying a permuted block design with random blocks stratified by children’s age. All patients who gave consent forms and who fulfilled the inclusion criteria were randomized. Randomization was requested by the staff member responsible for recruitment and clinical interviews and was performed by the computer technician, both members of NITIDA. Closed envelopes with printed randomized numbers on them were available for the therapist, who was not involved in assessing the outcome of the study. Once the initial assessment was completed, the therapist verified the allocation, and the intervention was initiated. Staff responsible for recruitment and symptom ratings were not allowed to receive information about the group allocation.

The researchers responsible for data management and statistical analysis were blind to the research groups. Baseline comparisons were performed using the chi-square test for categorical data and one-way ANOVA for dimensional data. Our sample size offers 99% power to detect large (0.40) or moderate effect sizes (0.25) and 82% for small (0.15) effect sizes with a 5% error probability. We estimated the correlation between repeated measures using the test–retest reliability coefficient for the psychometric tests (an average of 0.7). The power analysis was conducted in the G*Power 3.1.9.7 software (Faul et al., 2007).

The comparison between pre- and post-intervention measures was performed using analysis of variance for repeated measures. Our sample had missing data post-intervention. Although there was little missing data, the sample available for each variable varied between 54 and 57, with an average of 56.4. A visual analysis of missing data and the MCAR test suggested a pattern in which the variables were missing completely at random, so we used multiple imputations using the automatic method (based on linear regression) of SPSS 25.0 (IBM Corp. Released, 2017). All available sociodemographic, baseline, and outcome measures were used as predictors. We computed 10 virtual datasets using this method and condensed them into a unique database for further analysis, containing the mean of each variable across the 10 imputations for each subject.

We specifically analyzed the effect of the interaction between the time factor (pre × post-intervention) and the group factor (ST × ST + Online PT × ST + Face-to-Face PT) in terms of statistical significance and effect size (using the partial eta-square). Age was added as a covariate. For this calculation, we typically considered a small effect size of 0.01, moderate values of 0.06, and large values of 0.15 (Hair et al., 2009). Post-hoc analyses were performed using the Bonferroni–Holm method for multiple comparisons. The latter procedures were performed in JASP 0.16.4 (JASP Team, 2022).

The results regarding the effect of the interventions on ADHD symptoms can be found in Figure 2. The comparisons between pre- and post-treatment were significant for all MTA-SNAP-IV measures, reflecting an important symptom reduction: inattention (F (54) =22.23, p < 0.001, η_p² = 0.29), hyperactivity-impulsivity (F (54) =19.79, p < 0.001, η_p² = 0.27), and ODD (F (54) =16.51, p < 0.001, η_p² = 0.13).

Regarding treatment outcomes, we found a significant interaction between timepoint (baseline × follow-up) and treatment modality (ST, ST + Online PT, ST + Face-to-Face PT) for MTA-SNAP-IV inattention (F (54) =3.72, p = 0.030, η_p² = 0.12) and MTA-Snap-IV ODD (F (54) =3.89, p = 0.026, η_p² = 0.13) but not for MTA-SNAP-IV hyperactivity (F (54) =2.90, p = 0.063). These results are summarized in Figure 2.

There were no group differences in baseline. Post-hoc analysis (corrected for multiple comparisons using the Bonferroni–Holm method) for MTA-SNAP-IV inattention suggests that ST alone did not show significant improvement between pre- and post-treatments (mean difference = 0.35, p = 1.000), but the addition of Face-to-Face PT does (mean difference = 3.24, p = 0.001). ST + Online TP showed a similar trend but did not reach statistical significance (mean difference = 2.11, p = 0.093). The comparison between the PT modalities showed no significant differences (mean difference = 1.12, p = 1.000).

For MTA-SNAP-IV ODD symptoms, ST alone also did not reduce symptoms between baseline and follow-up (mean difference = 0.050, p = 1.000), but the addition of Face-to-Face PT (mean difference = 1.49, p = 0.033) or online PT (mean difference = 1.65, p = 0.009) does it. There was no significant difference between PT modalities (mean difference = −0.16, p = 1.000).

Table 2 presents the results regarding the PSDQ measures. In addition to the subscales, support and affection (p = 0.081) showed significant differences between pre- and post-treatment [average effect size (η_p² = 0.13)]. These differences were all toward positive parenting (reduction of authoritarian and permissive methods and increase of authoritative ones). We found a significant interaction between pre-/post-intervention and treatment modality in PSDQ support and affection subscales—a component of the Authoritative Parental Style (p = 0.005), but post-hoc analyses were not significant after correcting multiple comparisons. The other results of PSDQ main scores or subscales did not reach statistical significance.

The results of this section are shown in Table 3. Our analysis suggests a significant improvement in the psychological, mood and emotion, parents, financial, and school aspects of the patient’s quality of life in the Kidscreen-52 (al p < 0.05, average effect size of of η_p² = 0.13) but non-significant results for physical, self-perception, autonomy, or peers. We found an interaction between pre−/post-treatment and treatment modality only in the physical domain of quality of life (F (54) =5.22, p = 0.009, η_p² = 0.16). Post-hoc analysis suggests improved quality of life when the ST was compared to both in ST + Online PT (mean difference = −5.90) and ST + the Face-to-Face PT (mean difference = −5.25), although these differences did not remain after adjusting for multiple comparisons (p = 0.444 and p = 0.556, respectively).

The results of this section are shown in Supplementary Table S1. Quality of life in the physical (F (47) =7.54, p = 0.009, η_p² = 0.14), psychological (F (52) =8.59, p = 0.005, η_p² = 0.14), and environmental domains (F (49) =8.93, p = 0.004, η_p² = 0.15) were higher at the post-treatment assessment when compared to pretreatment. No differences were observed in the social domain of quality of life or in the perceived stress scale. No interactions with treatment modality were significant in these analyses.