Family-Based Treatment of Early Childhood OCD: The Pediatric OCD Treatment Study Junior (POTS Jr.) Randomized Controlled Trial

Abstract

Importance.

Cognitive-behavioral therapy (CBT) has been established as efficacious for obsessive compulsive disorder (OCD) in older children and adolescents, yet its effect in young children has not been evaluated sufficiently.

Objective.

To examine the relative efficacy of family-based CBT (FB-CBT) involving exposure plus response prevention (EX/RP) versus a family-based relaxation treatment (FB-RT) control condition for children ages 5 – 8.

Design, Setting, and Participants.

A 14 week randomized controlled trial (RCT) conducted at three academic medical centers between 2006 and 2011, involving 127 pediatric outpatients ages 5 – 8 with a primary diagnosis of OCD and a Children’s Yale-Brown Obsessive Compulsive Scale (CY-BOCS) total score of 16 or higher.

Interventions.

Participants were randomly assigned to 14 weeks of: (1) FB-CBT including exposure plus response prevention (EX/RP); or (2) FB-RT.

Main Outcome Measures.

Responder status defined as an Independent-Evaluator (IE) rated Clinical Global Impression - Improvement (CGI-I) score of 1 (very much improved) or 2 (much improved) and change in IE-rated continuous CY-BOCS total score.

Results.

FB-CBT was superior to FB-RT on both primary outcome measures. The percentages of children who were rated as 1 (very much improved) or 2 (much improved) on the Clinical Global Impression-Improvement scale at 14 weeks were 72% for FB-CBT and 41% for FB-RT. The effect size difference between FB-CBT and FB-RT on the CGI-I was 0.31 (95% CI, 0.17 to 0.45). The number needed to treat (NNT) with FB-CBT versus FB-RT was estimated as 3.2 (95% CI, 5.8 to 2.2). The effect size difference between FB-CBT and FB-RT on the CY-BOCS at Week 14 was 0.84 (95% CI, 0.62 to 1.06).

Conclusions and Relevance.

A comprehensive, family based CBT program was superior to a relaxation program with similar format in reducing OCD symptoms and functional impairment in young children (ages 5-8) with OCD.

Early childhood onset obsessive compulsive disorder (OCD) disrupts social, family, and academic functioning, compromising achievement of normal developmental milestones^1-4. The efficacy of cognitive-behavioral therapy (CBT), selective serotonin inhibitors, and their combination has been established for older children and adolescents with OCD^5-10, yet little is known about their efficacy in younger children. Young children with OCD have been found to have similar clinical profiles, including comparable obsession/compulsion types and multiple comorbidities^11-15. Only religious/scrupulosity obsessions and an increased likelihood of depressive disorders are more common in older children with OCD^11,12,14. Because pediatric OCD’s pernicious impact on functioning extends into adulthood^16-18, developing effective, developmentally sensitive interventions for early-emerging OCD is a public health imperative.

Contemporary cognitive behavioral therapies (CBT) for older children do not adequately address the unique features of OCD in young children (ages 5-8), especially: developmental differences, family context, unique symptom correlates, and family’s initial contact with the mental health system. Although CBT for older youth allows for parent involvement, it often does not provide explicit, systematic instructions for structuring parental participation. It also does not provide specific strategies for explaining concepts to patients with less advanced cognitive abilities.

An evaluation of our developmentally sensitive, family-based CBT protocol (FB-CBT) adapted for youth ages 5 – 8 provided promising results^19,20. Adaptations addressed cognitive, socio-emotional, and family contextual differences for young children, while maintaining emphasis on exposure plus response prevention (EX/RP), the CBT component with the most empirical support for treating OCD²¹. A small randomized controlled trial (RCT) yielded moderate and large treatment effects for FB-CBT for intent-to-treat (d = 0.53) and completer (d = 0.85) samples, respectively, when compared to family-based relaxation treatment (FB-RT), a credible psychosocial control condition²⁰. Demonstrating acute efficacy definitively, however, requires a larger sample, multiple sites to permit examination of generalizability, and a broader evaluation of change across OCD symptoms, functional impairment, and quality of life.

Toward these ends, our collective research group, which has already examined the efficacy of CBT, pharmacotherapy, and their combination⁹ as well CBT’s efficacy in augmenting serotonin-reuptake inhibitor (SRI) partial response¹⁰, initiated the Pediatric OCD Treatment Study Junior, or POTS Jr. Study. In the present study, we hypothesized that FB-CBT would yield a greater response rate and improvements on continuous measures of OCD and related dysfunction compared to a relaxation protocol with similar format and developmental considerations.

METHODS

Design

POTS Jr. was a 14 week, parallel group controlled trial that involved three sites (Brown University, University of Pennsylvania, and Duke University) and two treatment conditions (FB-CBT vs. FB-RT). Acute phase outcomes were measured at the Week 0 baseline visit and again at Weeks 5, 9, and 14. Detailed descriptions of design considerations, measurement psychometrics, recruitment procedures, and adaptations designed to increase the developmental sensitivity of the assessment and treatment are published elsewhere⁴. A summary of the trial protocol is available at www.clinicaltrials.gov, number NCT00533806 and the full protocol is available from the author. Each site’s institutional review board approved the study.

Participants

Inclusion criteria were: (1) ages 5-8; (2) primary OCD according to the Diagnostic and Statistical Manual of Mental Disorders (Fourth Edition, Text Revision) (DSM-IV-TR)²²; (3) clinically relevant OCD symptoms defined as ≥16 on the Children’s Yale-Brown Obsessive-Compulsive Scale (CY-BOCS)²³; (4) stable symptoms for ≥ 3 months; and (5) outpatient. Exclusion criteria were: (1) pervasive developmental disorder(s); (2) pediatric autoimmune neuropsychiatric disorders associated with strep infection (PANDAS); (3) concurrent psychotherapy; (4) acute suicidality; (5) concomitant medications not stable for > 8 weeks; or (6) prior trial of CBT for OCD. The “primary” diagnosis was defined as the disorder causing the most significant functional impairment and clinical distress based on the KSADS diagnostic interview with the parent(s) and child; participants could have other co-primary diagnoses and still be included.

Given that children with streptococcal precipitated OCD may have a more episodic symptom course, we considered carefully the implications of including children with PANDAS in the current study. Our decisions were guided by the state of the PANDAS literature in 2006 when the study was funded. Participants whose parents reported during the initial phone screening that the child had been previously diagnosed with PANDAS or who upon in-person assessment met the published research diagnostic criteria for PANDAS ²⁴ and/or were taking antibiotics for treatment/prophylaxis of OCD, tic symptoms, or rheumatic fever, were excluded.

Determination of Eligibility

Eligibility was determined via a three-gate procedure utilized in our prior RCTs^9,10. Gate A included a brief screening interview with a parent/guardian (telephone and/or in-person). The Gate B intake interview included: 1) obtaining informed consent, 2) assessment of OCD symptomology, and 3) assessment of comorbidity. Gate C included the Week 0 baseline assessment, followed by a separate meeting that revealed the randomization outcome to the family. Progression from Gate A through C was typically completed within three weeks and is summarized in the Consort Diagram (Figure 1).

Figure 1.

CONSORT DIAGRAM

Randomization

Patients were randomly assigned to one of the two treatment conditions between 2006 and 2011 using a computer-generated permuted blocking procedure, stratified by site, SRI medication status, and presence of co-morbid tics at baseline.

Treatments

Treatment delivery schedule

Twelve sessions were delivered over 14 weeks in both conditions. The first two sessions (90 minutes) were conducted with parents only, while remaining sessions (60 minutes) were conducted jointly with parents and children. The last two sessions were held over four weeks.

Outline of Treatment Components

FB-CBT focused on providing child and parent “tools” to understand, manage, and reduce OCD symptoms. Primary components included: 1) psychoeducation, 2) behavior management skills training (parent tools), 3) externalizing OCD and EX/RP (child tools), and 4) family process components. Psychoeducation topics included OCD’s neurobiology, correction of OCD misattributions, identifying OCD behaviors, and rationale for treatment. Parenting tools included behavioral management of the child’s OCD symptoms with differential attention (e.g., ignoring, rewards), modeling, and scaffolding (i.e., parent guides child’s use of tools, gradually giving the child less help to promote self-regulation). All parenting tools were rehearsed in session and practiced at home as part of weekly homework assignments. EX/RP emphasized active collaboration with the family by assisting in hierarchy development and implementing gradual exposure to triggers. Parents were actively involved during in-session and home-based EX/RP.

Key adaptations to CBT to better fit this young age group included: modified psychoeducation, increased focus on parent-based skills, and simplification of CBT skills. The following adaptations to standard individual OCD treatment models were most crucial: 1) involvement of parents in all phases of treatment; 2) tailoring of psychoeducation, exposures, and homework to meet the child’s unique developmental level; and 3) focus on the family context (e.g., lack of familiarity with the mental health system, patterns of family accommodation, parental psychopathology) and in particular, the parent’s response to their child’s anxieties.

FB-RT focused on implementing relaxation strategies aimed at lowering the child’s anxiety. FB-RT components included: 1) psychoeducation, 2) affective education, and 3) relaxation training. Psychoeducation content included the relationship between stress management and anxiety, rationale for treatment, and implementing a reward system devoid of other differential attention principles. Affective education emphasized teaching the child how to identify negative and positive feelings with special emphasis on recognizing anxiety. Relaxation training consisted of developmentally adapted instruction in progressive muscle relaxation and guided imagery.

Pharmacotherapists provided oversight for patients on a stable SRI medication regimen at each IE visit. Side effects, with a special focus on suicidal ideation and/or behavior, were closely monitored, but families did not transfer medication management to the study physician.

Supervision and Oversight

Each site had major supervisory responsibilities: Brown provided clinical supervision of FB-CBT and FB-RT; Duke organized data management and statistical analyses; and Penn provided supervision of diagnostic interviews and independent evaluator (IE) assessments. Treatment providers for FB-CBT and FB-RT were clinical psychologists and clinical psychology trainees already familiar with CBT. Therapists completed in-person training at study initiation and extensive site- and cross-site supervision in both treatment arms. To assess treatment fidelity, approximately 12% of intervention sessions (n = 92 sessions per treatment) were coded independently by trained raters. Sessions were rated on adherence to session-specific content, and each had prescribed and proscribed targets. Each target was rated between 1 (none to low), 2 (acceptable), and 3 (excellent) adherence. Ratings indicated high adherence to both prescribed elements (FB-CBT = 2.86 (0.09); FB-RT = 2.94 (0.07)) and proscribed elements (FB-CBT = 2.96 (0.07); FB-RT = 3.00 (.01)). An independent data and safety monitoring board provided regular oversight and met bi-annually during the trial.

Assessment

IEs were all doctoral level psychologists and blind to treatment condition; they were trained to reliability on the CY-BOCS and Clinical Global Impression-Severity (CGI-S) and Improvement scales (CGI-I)²⁵ through joint interviews, videotape reviews, and participation in monthly cross-site supervision conferences. Reliability was checked regularly on randomly selected tapes, and IEs were retrained if they fell below 80% agreement. IEs were trained to adapt their interview procedures as needed to maximize developmental sensitivity of assessments⁴.

IE evaluations were completed at Weeks 0, 5, 9 and 14; all parent- and therapist-report measures were completed on scheduled visit days. Consistent with an intent-to-treat approach, all patients were assessed at each time point.

Measures

OCD Symptoms, Severity, and Improvement were measured using the CY-BOCS, CGI-S, and the CGI-I. The CY-BOCS is a clinician interview that merges data from clinical observation and parent and child report. For the purposes of this study, all CY-BOCS interviews were completed with parent and child together. The literature supports the use of the measure in children as young as 6 years²⁶ and it was used successfully in our prior work^11,20 with 5 year olds. The CGI-S is a 7-point scale measuring clinician-rated illness severity. The CGI-I is a 7-point scale measuring clinician-rated improvement in treatment^27,28. Those who were rated as “Much Improved” or “Very Much Improved” on the CGI-I were considered as meeting the criterion for RESPONSE (i.e., the primary dichotomous outcome measure). The CGI has been used in multiple child psychiatric studies with children as young as 40 months²⁹ and in children as young as 3 years old with OCD^11,20,30,31.

Functioning was measured using the Children’s OCD Impact Scale-Revised (COIS-R³²), which provides a standardized format for assessing OCD’s impact on social, school, and home functioning in children ages 5-17³³. The parent report form of the measure was used for this study.

Quality of Life was measured using the Pediatric Quality of Life Enjoyment and Satisfaction Questionnaire (PQ-LES-Q). The PQ-LES-Q is a 15-item parent-report scale measuring QOL. The scale has solid psychometric properties with excellent internal consistency and adequate concurrent validity and was developed for children ages 6-17³⁴. We used a parent report of this measure, based on the finding that close relatives are able to give accurate proxy ratings about QOL³⁵.

Demographics were measured using the caregiver-reported Conners-March Developmental Questionnaire (CMDQ³⁶), including age, grade level, gender, race, and socioeconomic status.

Comorbid psychiatric disorders were assessed with the Kiddie Schedule for Affective Disorders and Schizophrenia for School Age Children-Present and Lifetime Version (K-SADS-P/L)^37,38, which is a semi-structured, clinician interview that yields DSM-IV-TR diagnoses across Axis I domains. Interviews were administered to the primary caregiver(s) regarding the child, and to children (although 5-6 year old children varied in their ability to participate actively). The K-SADS-P/L is routinely used to assess psychiatric diagnoses in children as young as 5 years^39,40.

Sample Size and Power

Sample size determination was based on response rates of 45% for FB-CBT vs. 20% for FB-RT taken from our prior pilot study²⁰. For the simplest versions of the primary aims, a comparison of two independent binomial proportions using Pearson's Chi-square statistic with a Chi-square approximation with a two-sided significance level of 0.05, group sample sizes of 62 and 62 (total N = 124) have an approximate power of 0.853 when the proportions are 0.45 and 0.20.

Missing Data

As part of the study design, efforts were made to collect all outcomes on all randomized participants even when treatment was prematurely terminated⁴¹. Prior to analysis, we used multiple imputation to replace missing values⁴². A sequential regression multivariate imputation algorithm⁴³ implemented in the IVEware package for SAS⁴⁴. The imputation model included all longitudinal outcome measures, time since randomization, treatment indicators, putative moderators and mediators⁴¹, and the baseline stratification variables. Five data sets were generated. Results reported below were calculated using Rubin’s rules⁴³ for combining the results of identical analyses performed on each of the five imputed data sets.

Statistical Analysis

All randomized participants were included in the analyses, in accordance with intention-to-treat principles. A multivariate chi-square test was used to test for between-group differences in RESPONSE rates at Week 14. Separate longitudinal regression models were used to examine mean differences in the two continuous outcomes (CY-BOCS, COIS-R) between conditions at each assessment visit. Each regression model included indicators of time (assessment visit), group assignment, and all time-by-group interaction terms. Baseline stratification variables employed in the randomization procedure (site, SRI medication status, and tics) were also included in each model a priori. Residual error terms were assumed to follow a mean-zero, normal distribution with an unstructured covariance structure used to capture the within person correlation over time. The fitted models reported mean scores at each assessment visit and make inferences about between-groups comparisons at the final assessment visit. Tests were two-sided, and a P value of less than 0.05 was considered statistically significant. The sequential Dunnett test was used to control the overall (familywise) error rate⁴⁵. Longitudinal models were fit using PROC MIXED in SAS Statistical Software, Version 9.2 Level 2M2 (SAS Institute, Cary, NC). Throughout the analyses, adjusted degrees of freedom are implemented and presented using the EDF option in SAS. In our case, we used the SAS macro COMBCHI⁴⁶.

To enhance interpretation of the results for RESPONSE, we calculated the number needed to treat (NNT)⁴² with FB-CBT relative to FB-RT. For the continuous outcomes, we calculated standardized between-group mean differences⁴⁷ at the Week 14 visit.

RESULTS

Recruitment and Retention

The Consort Diagram is depicted in Figure 1. Recruitment took place between 2006 and 2011 and ended once the recruitment goal was met. The final follow-up visit was completed in 2013. Participants were recruited from: (1) site clinics; (2) schools; (3) primary care physicians; (4) mental health providers; and (5) paid and public service advertisements in local media. Of the 127 subjects who underwent randomization, 126 (99.2%) completed at least one post-baseline assessment. The mean number of completed sessions in FB-CBT was 11.2 (CI 95%, 11.8 to 10.6) out of a possible 12 sessions. The mean number of completed sessions in FB-RT was 10.1 (CI 95%, 10.87 to 9.31). A total of 102 (80.3%) participants completed acute treatment: 8 dropped out of the study and were lost to follow-up (FB-RT = 6, FB-CBT = 2), 7 dropped out of treatment but remained in the study for outcome assessments (FB-RT=2, FB-CBT=5), and 10 prematurely terminated the assigned treatment due to lack of efficacy, received out-of-protocol treatment, but remained in the study for outcome assessments (FB-RT = 9, FB-CBT = 1). Investigation of post-randomization activity indicated that FB-RT patients were more likely to prematurely terminate and receive out-of-protocol treatment (χ2(1) 4.75, p < .03). There was one adverse event reported in the study. This patient was prematurely terminated from FB-CBT in order to seek additional psychopharmacological treatment for aggressive, impulsive behavior.

Sample Characteristics

Table 1 summarizes the baseline and clinical characteristics presented by treatment condition. No significant between groups differences emerged. Few participants (7.1%) were on psychotropic medication during the trial and were equitably distributed between treatment arms.

Table 1.

Baseline Characteristics and Observed Cases by Timepoint

VARIABLE	FB-RT (n = 64)	FB-CBT (n = 63)	All Subjects (n = 127)
STUDY CENTER – N (%)
Brown University	25 (39.1)	23 (36.5)	48 (37.8)
Duke University Medical Center	17 (26.6)	18 (28.6)	35 (27.6)
University of Pennsylvania	22 (34.4)	22 (34.9)	44 (34.7)
DEMOGRAPHICS
Age (SD), years	7.04 (1.2)	7.40 (1.2)	7.22 (1.2)
Age of Onset	4.68 (1.6)	5.38 (1.7)	5.08 (1.7)
5-6 y –N (%)	31 (48.4)	23 (36.5)	54 (42.5)
Female –N (%)	28 (43.8)	39 (61.9)	67 (52.8)
Race –N (%)
White	55 (85.94)	59 (93.65)	114 (89.76)
Black	1 (1.56)	1 (1.59)	2 (1.57)
Asian	3 (4.69)	0	3 (2.36)
Mixed	3 (4.69)	1 (1.59)	4 (3.15)
Not Reported	2 (3.13)	2 (3.17)	4 (3.15)
Ethnicity
Not Hispanic/Latino	61 (95.31)	60 (95.24)	121 (95.3)
Hispanic/Latino	3 (4.69)	3 (4.76)	6 (4.7)
Mean Family Income (SD)	70k-80k (20k-30k)	80k-90k (20k-30k)	70k-80k (20k-30k)
OCD BASELINE SEVERITY
CY-BOCS (SD)	25.97 (3.98)	25.13 (4.46)	25.55 (4.23)
COIS-R (SD)	23.46 (12.68)	23.97 (16.43)	23.72 (14.62)
CGI-S (SD)	4.67 (0.76)	4.71 (0.89)	4.69 (0.82)
CGI-S<5 –N (%)	30 (46.88)	26 (41.27)	56 (44.09)
CGI-S≥5 –N (%)	34 (53.13)	37 (58.73)	71 (55.91)
BASELINE COMORBIDITIES
Any –N (%)	35 (54.69)	40 (63.49)	75 (59.06)
Anxiety –N (%)	29 (45.31)	30 (47.62)	59 (46.46)
Separation Anxiety	8 (12.50)	8 (12.70)	16 (12.60)
Specific Phobia	11 (17.19)	16 (25.40)	27 (21.26)
Social Phobia	10 (15.63)	4 (6.35)	14 (11.02)
Generalized Anxiety Disorder	13 (20.31)	12 (19.05)	25 (18.69)
Mood – N (%)	1 (1.56)	1 (1.59)	2 (1.57)
Dysthymia	1 (1.56)	0	1 (0.79)
Depressive Disorder NOS	0	1 (1.59)	1 (0.79)
Tic Disorder – N (%)	16 (25.00)	13 (20.63)	29 (22.83)
Tourette Syndrome	7 (10.93)	7 (11.11)	14 (11.02)
Chronic Motor Tic Disorder	3 (4.69)	2 (3.17)	5 (3.15)
Chronic Vocal Tic Disorder	3 (4.69)	1 (1.59)	4 (1.57)
Transient Tic Disorder	3 (4.69)	3 (4.76)	6 (4.72)
Past Tics Reported	1 (1.56)	2 (3.17)	3 (2.36)
Externalizing –N (%)	13 (20.31)	18 (28.57)	31 (24.41)
ADHD	10 (15.63)	8 (12.70)	18 (14.17)
Oppositional Defiant Disorder	6 (9.38)	12 (19.05)	18 (14.17)
Elimination Disorder –N (%)	5 (7.81)	2 (3.17)	7 (5.51)
Enuresis	5 (7.81)	2 (3.17)	7 (5.51)
Encopresis	1 (1.56)	0	1 (0.79)
PSYCHOTROPIC MEDICATION
ADHD –(N) (%)	4 (6.25)	3 (4.76)	7 (5.51)
SRI –(N) (%)	0	2 (3.17)	2 (1.57)
Other –(N) (%)	0	1 (1.59)	1 (0.79)
OBSERVED CASES – N
Baseline	64	63	127
Week 5	62	62	124
Week 9	57	58	115
Week 14	57	59	116

Abbreviations: ADHD: Attention-Deficit/Hyperactivity Disorder; CGI-S: Clinical Global Impression Scale – Severity; COIS-R: Child Obsessive-Compulsive Impact Scale-Revised; CY-BOCS, Children’s Yale-Brown Obsessive Compulsive Scale; FB-RT: Family-Based Relaxation Therapy; FB-CBT: Family-Based Cognitive Behavioral Therapy; SRI: Serotonin Reuptake Inhibitor

Clinical Response and Effect Sizes of Clinical Significance

Using site, gender, baseline age (in months), baseline CGI-S, medication status (coded as taking SRI medication at baseline or not), and tic disorder as covariate(s), the percentages of children who were rated as 1 (very much improved) or 2 (much improved) on the CGI-I scale at 14 weeks were 72% for FB-CBT and 41% for FB-RT. The number needed to treat (NNT) with FB-CBT versus FB-RT to see one additional responder at Week 14, on average, was estimated as 3.2 (95% CI, 5.8 to 2.2). On continuous measures with the same covariates included in the model, FB-CBT was superior to FB-RT at Week 14 on the CY-BOCS (see Figure 2) and COIS-R. Treatment effect sizes between groups for Week 14 CY-BOCS and COIS-R were 0.84 (95% CI, 0.62 to 1.06) and 0.42 (95% CI, 0.06 to 0.77), which corresponds to a large and medium standardized effect size, respectively. Random regression analyses identified a statistically significant linear trend with time (t = −18.14, p<.001) and a time-by-treatment interaction (t=−6.54, p<.001). There were no significant differences between conditions at Week 14 on the PQ-LES, as the 95% CI included zero (0.23; 95% CI, −0.09 to 0.55). Table 2 provides point estimates, planned comparisons, and the respective effect sizes on each continuous variable.²⁴

Figure 2.

Results of Fitting Longitudinal Model

Table 2.

Group-specific response rates, mean scores, and between-group effect sizes at Week 14.

Week 12 Variable	Responder Statusa	CY-BOCSb (95% CI)	COIS-Rc (95% CI)	PQ-LES-Qd (95% CI)
Estimated Meanse
Family-Based CBT	0.72 (0.58, 0.83)	12.30 (10.95, 13.65)	11.68 (8.80, 14.56)	4.16 (4.02, 4.29)
Family-Based RT	0.41 (0.28, 0.55)	19.67 (18.31, 21.03)	16.52 (13.62, 19.41)	4.02 (3.88, 4.15)
Effect Sizesf
Family-Based CBT v. Family-Based RT	0.31 (0.17, 0.45)	0.84 (0.62, 1.06)	0.42 (0.06, 0.77)	0.23 (−0.09, 0.55)
Family-Based CBT NNT	3.2 (5.8, 2.2)

Abbreviations: CBT: Cognitive Behavioral Therapy; COIS-R: Child Obsessive-Compulsive Impact Scale-Revised; CY-BOCS, Children’s Yale-Brown Obsessive Compulsive Scale; NNT: Number Needed to Treat; PQ-LES-Q: Pediatric Quality of Life Enjoyment and Satisfaction Questionnaire; RT= Relaxation Therapy.

^aResponder Status scores range from 0.0 to 1.00 reflecting the percentage of responders who were rated as Much Improved or Very Much Improved on the Clinical Global Impression-Improvement Scale.

^bCY-BOCS scores range from 0 to 40 with larger scores reflecting more OCD symptoms.

^cCOIS-R is a 33-item rating scale with items scored from 0 (not at all) to 3 (very much), with higher scores indicating greater functional impairment caused by OCD symptoms.

^dPQ-LES-Q is a 15-item rating scale with items scored from 1 (very poor) to (very good); the first 14 items are summed, with higher scores reflecting greater enjoyment and satisfaction

^eEstimated means were modeled using site, gender, baseline age (in months), baseline severity (CGI-S), medication status (taking medication at baseline or not), and presence of tic disorder at baseline as covariates.

^fFor Responder Status: between-groups difference in estimated response rate at Week 14 (95% CI). For CY-BOCS, NIMH, and CGI-S: between-groups difference in estimated mean score at Week 14 divided by the pooled standard deviation of the outcome at Week 14, otherwise known as Cohen’s d (95% CI). The magnitudes associated with Cohen’s d have recognized conventions (i.e., Small effect = 0.20-0.49; Medium effect = 0.50-0.79; Large effect = > 0.80).¹ All effect size estimates are reported such that positive scores indicate that the first treatment group was superior to the comparison group in functionin

¹Cohen J. Statistical power analysis for the behavioral sciences. 2nd ed. Hillsdale, NJ: L. Erlbaum Associates; 1988.

A multivariate chi-square test found no statistically significant site x treatment interactions for RESPONSE at the Week 14 visit (p = .48). Similarly, no time x site x treatment interactions were found at Week 14 visit for the continuous outcomes: CY-BOCS (p = .58), COIS-R (p = .31), or PQ-LES (p = .82).

COMMENT

Our developmentally-sensitive, family-based CBT program that included EX/RP was found more efficacious in reducing OCD symptoms and functional impairment in young children (ages 5-8) than a similarly structured relaxation program. Considering the general absence of knowledge about the efficacy of treatments for young children with early emerging OCD, these findings have significant public health implications. These findings are also consistent with those of an earlier randomized pilot study²⁰ as well as several studies of family-based treatment for OCD in older children^5,8.

Given the use of an active control treatment in this trial, the magnitude of the difference in response rates between groups is particularly notable⁴⁸. FB-RT was equivalent to FB-CBT in the amount of therapist contact time and included the cogent and face-valid rationale of using anxiety management strategies while also controlling for treatment expectancy, acceptability, credibility, and the effects of repeated assessment. Response rates on the CGI-I in this trial (72% for FB-CBT) were even better than in our group’s pilot study²⁰ with the same population (50% for FB-CBT) and similar or better to our other multisite OCD studies: POTS I (75% for combined CBT and SSRI; 64% for CBT alone) and POTS-II (65.7% for combined CBT and Medication Management), which may reflect the importance of providing treatment closer in time to when the OCD first emerges. A particular strength of these comparisons is the use of a common metric to define response. Naturally, this is only one definition of treatment response, yet we see a larger effect size difference on the continuous measure of outcome.

FB-CBT was not only superior in reducing OCD symptoms on both categorical and continuous measures of outcome, but also reduced OCD-related dysfunction more than did the control condition. The effect on OCD related dysfunction, though significant, was not as robust as the effect on OCD symptoms. Some previous trials found that pediatric OCD symptom improvements preceded improvement in functioning⁴⁹, while others did not⁷. In this early age, perhaps the impact on functioning may take longer to observe, as a young child gradually approaches situations more appropriately only after significant symptom reduction.

Contrary to our hypotheses, FB-CBT was not significantly superior to FB-RT on Quality of Life ratings. Quality of Life has been shown to significantly improve for adolescents undergoing CBT treatment for OCD⁷, but perhaps for younger children improvements in the broader construct of quality of life may lag behind improvements in symptoms and functioning. It is possible that assessment of the construct of quality of life in these very young children may require different questions/content. The absence of change in quality of life in our study also may have been due to the fact that the PQ-LES-Q, the measure we used, was not exclusively designed for younger children (despite being validated for ages 6-17) and that scores were not in the clinical range at baseline.

The clinical implications of our results highlight the use of this family-based CBT model as the first line choice for young children with OCD. Despite little psychotropic medication use in this sample, children with early onset OCD are indeed able to benefit from a treatment approach that is uniquely tailored to their developmental needs and family context. Family-based EX/RP treatment is effective, tolerable and acceptable to young children and their families, and appears to be feasible to implement without concomitant pharmacotherapy. The sample included children with a range of OCD severity (although as a group they were in the moderate-severe range) and co-occurring psychiatric conditions, which enhances generalizability of findings to clinical samples of young children with OCD.

The study has several limitations that warrant mention. First, despite following strategies recommended by expert consultants involved in the study, recruitment paralleled the broader OCD treatment literature with regard to limited enrollment of racial and ethnic minority families and youth. This is a significant weakness and leaves the applicability of our findings to a broader range of ethnic and racial groups in question. Similarly, the lack of socioeconomic diversity in the sample is a concern. Another potential threat to generalizability lies in the high level of expertise at these respective sites and, by extension, of the clinicians who provided the treatments. Preliminary evidence from open studies of CBT (e.g.,^50,51) support its effectiveness for pediatric OCD in community clinics with patients seven and older, but whether these findings are applicable to these community contexts with younger children remains to be discovered. Questions remain as well about the generalizability of outcomes achieved with families willing to accept randomization, although a large CBT effectiveness study in adult OCD found outcomes with non-randomized OCD patients slightly better than those reported from randomized trials⁵². Finally, albeit beyond the scope of this paper, the durability of gains on primary outcomes and on longer-term functioning will be examined separately, as will moderators and mediators of treatment outcome.

In sum, our findings add to the evidence base supporting CBT for pediatric OCD by extending downward the age range that can benefit from CBT protocols emphasizing EX/RP. With appropriate parental support, young children with OCD can make significant gains in FB-CBT beyond what can be expected from having parents attempt to teach relaxation strategies to their children with OCD. Hopefully these gains can decrease the chronicity and morbidity of this debilitating illness on overall development. Our findings lend further support that these young children have “real” and impairing OCD that warrants more than a “watch and wait” approach. Finally, this study underscores the need to disseminate these treatment models beyond academic medical centers to clinical settings (e.g., pediatric offices, community clinics) where young children first present for treatment.