Participants were 80 individuals who completed a long-term follow-up evaluation (34) 11.17 years (SD = 1.25 years) after completion (i.e., 10-week post-treatment timepoint) of the original randomized controlled trial examining the efficacy of CBIT relative to Psychoeducation and supportive therapy [PST]) in 126 children and adolescents (9). See Table 1 and Results for sample characteristics. Among the long-term follow-up participants, 38 were originally randomized to CBIT and 42 were randomized to PST. Among those who received CBIT, 55.6% (n = 21) were classified as treatment responders and 44.7% (n = 17) were non-responders, and among those who received PST, 14.3% (n = 6) were classified as treatment responders and 85.7% (n = 36) were non-treatment responders (See Procedure and Data Analysis section for definition of treatment responder and 34 for more details regarding sample characteristics).

Inclusion criteria for the original trial included an age of 9 to 17 years; diagnosis of Tourette’s disorder, chronic motor tic disorder, or chronic vocal tic disorder, moderate or greater severity as measured by a Yale Global Tic Severity Scale (YGTSS; 37) Total score >13 (>9 for chronic motor tic disorder or chronic vocal tic disorder); fluency in English; and an intelligence quotient > 80. Exclusion criteria included the need for immediate treatment or change in current treatment for any of the co-occurring psychiatric disorders allowed at study entry: ADHD, obsessive–compulsive disorder, anxiety disorders, depressive disorders, or oppositional defiant disorder; changes in the dosage or schedule of any psychotropic medications within six weeks of study enrollment or planned changes or initiation during the study; unstable medical condition; current diagnosis of substance abuse/dependence; lifetime diagnosis of pervasive developmental disorder, mania or psychosis; or ≥ 4 sessions of habit reversal training (9). Of the original CBIT sample (n = 126), 30 were lost to follow and 16 declined to participate. See Table 1 for baseline and post-treatment (10-week) demographic and clinical characteristics for participants who completed long-term follow-up (n = 80) and participants who were lost to follow up or declined to participate (n = 46). At follow-up, the sample ranged in age from 16 to 30 years (M = 22.87, SD = 2.70), was predominantly male (n = 60, 75.0%) and of non-Hispanic/Latino (n = 73, 91.2%) ethnicity. Participants endorsed white (n = 69, 86.3%), Black (n = 1, 1.3%), Asian (n = 4, 5.0%), multi-racial (n = 4, 5.0%), and other racial (n = 2, 2.5%) backgrounds. The majority of the sample were single/never married (n = 66, 82.5%) and had attained some college education or higher (n = 57, 71.3%). Half of the sample were employed (n = 40, 50.0%). See Espil et al. (34) for further methodological details.

Pittsburgh Sleep Quality Index (PSQI). The PSQI (61), administered at long-term follow-up is a 19-item self-report measure of sleep quality and disturbance over the previous month. Items are summed to produce seven subscale scores (sleep duration, sleep disturbance, sleep latency, daytime dysfunction due to sleepiness, sleep efficiency, overall sleep quality, and needs medication to sleep) are calculated from these items. These subscale scores are summed to yield a global sleep quality score. Higher scores indicate greater sleep disturbance. A PSQI global score greater than 5 is indicative of poor sleep quality (61). Forty-six participants in the long-term follow-up sample (57.5%) were above the cutoff for poor sleep quality. The PSQI has demonstrated acceptable reliability and validity (62–64). The internal consistency for PSQI Total in the present sample is.49.

See Table 2 for detailed descriptions of measures evaluating predictor variables. including demographics (i.e., age, sex) and medical history (i.e., tic medication and stimulant medication status), tic severity and tic-related impairment, anxiety severity, depression severity, and ADHD symptom severity.

Original CBIT Trial: Following institutional review board parent permission, child assent, and screening assessment procedures, participants and their parents completed a baseline evaluation for a 10-week randomized, controlled clinical trial evaluating the efficacy of CBIT relative to PST (9). CBIT is an 8-session blended intervention delivered over 10 weeks, involving habit reversal training, function-based assessment and intervention, relaxation training, and behavioral rewards (65). PST is a structured intervention providing supportive therapy and psychoeducation about tics developed to control attention and time (9). Treatment was delivered at three sites (University of California, Los Angeles [n = 45], University of Wisconsin-Milwaukee/Marquette University [n = 40] and Johns Hopkins University [n = 41]). This was followed a post-treatment evaluation during which a trained independent evaluator (IE) assessed tic severity and tic-related impairment (Yale Global Tic Severity Scale; YGTSS; 37) and rated degree of participant improvement in global illness-related functioning using the Clinical Global Impressions-Improvement Scale [CGI-I]; 66). Parents provided demographics and medical history and rated child behavior, including ADHD symptom severity (ADHD Rating Scale-IV; ADHD-RS-IV; 53, 54). Children rated anxiety severity (Screen for Child Anxiety Related Emotional Disorders; SCARED; 40) and depressive symptom severity (Children’s Depression Inventory; CDI; 45, 46).

Long-term Follow-up Assessment: At long-term follow-up assessment, participants from the original trial were recruited from two of the three original CBIT sites: University of California, Los Angeles (n = 32), Marquette University (n = 22), and Weill Cornell University (n = 26) due to the move of the original Johns Hopkins University principal investigator to that university. Adults provided institutional review board-approved informed consent for their participation. The three child participants (aged 16–17 years) provided assent, and their parents provided parent permission. The follow-up evaluation was completed in person or via web-based videoconferencing as needed. Independent evaluators (IEs) with a bachelor’s degree or higher, masked to original treatment assignment, and trained to reliability on clinical interviews (i.e., received didactic training and achieved reliable ratings on two consecutive interview administrations; see 34), evaluated psychiatric diagnosis (Mini International Neuropsychiatric Interview; MINI; 67), tic severity, and tic-related impairment (YGTSS). Participants rated sleep (PSQI Total), anxiety severity (An Adult Version of the Screen for Child Anxiety Related Disorders; SCARED-Adult; 44), depressive symptom severity (Mood and Feelings Questionnaire – Short Version; MFQ; 50), and ADHD symptom severity (Adult ADHD Self-Report Scale; ASRS; 57).

Analyses were conducted using SPSS 28.0. Descriptive statistics, including means and frequencies, are presented to characterize the sample with respect to demographics. Multiple linear regression was performed to examine whether post-treatment (i.e., 10-week timepoint of original CBIT trial) variables (including sex, tic medication status, stimulant medication status, YGTSS Tic Severity Total, YGTSS Impairment, SCARED Total, CDI Total, and ADHD-RS-IV Total) were predictors of the PSQI Total score at long-term follow-up. Multiple linear regression was also performed to examine whether variables measured at long-term follow-up (sex, tic medication status since the trial ended, stimulant medication status since the trial ended, YGTSS Tic Severity Total, YGTSS Tic-related Impairment Total, SCARED-Adult Total, CDI Total, and ADHD-RS-IV Total) were significant predictors of PSQI Total score. In both analyses, treatment assignment (CBIT versus PST), and treatment responder status (i.e., CGI-I rating of ‘much improved’ or ‘very much improved’ relative to a rating of ‘improved,’ ‘minimally improved,’ ‘no change,’ ‘minimally worse,’ ‘much worse,’ or ‘very much worse’) were included as covariates to account for prior tic treatment effects.

Then these analyses were repeated using multiple linear regression with backward elimination. Covariates (treatment assignment, treatment responder status) were entered into the model at each step of the backward elimination process, while predictor variables with a p-value of <.05 were removed from the model at each step in ordered fashion. Model fit was assessed at each step until achieving the final reduced model. Final reduced models were selected based on parsimony (i.e., simplest model that maintains goodness-of-fit) (36). We report adjusted R², indicating the proportion of variance explained by the model adjusted for the number of predictors in the model. A higher R² value indicates greater goodness-of-fit based on the predictors in the model (68). The variance-inflation factor (VIF) is reported as a measure of multicollinearity. Lower VIF scores indicate reduced collinearity among predictor variables. VIF scores above 10 indicate substantial multicollinearity among independent variables (69).

There were no significant differences in demographic and clinical characteristics (i.e., age, sex, tic medication status, stimulant medication status, YGTSS Total Tic Severity and Tic-related Impairment scores, SCARED Total, CDI Total, ADHD-RS-IV Total) at baseline of the original CBIT trial between participants who completed long-term follow-up and participants who were lost to follow-up or declined to participate (p = >.999 –.288). There were no significant differences in clinical chacteristics at post-treatment (10-week) timepoint of the original CBIT trail between participants who completed long-term follow-up and participants who were lost to follow up or declined to participate (p = .644 –.873). For bivariate correlations among clinical measures at post-treatment (10-week) and long-term follow-up see Supplementary Tables 1, 2.

In the full model, there were no significant post-treatment (10-week) predictors of PSQI Total at long-term follow up (see Table 3 and Supplementary Table 3). The full model explained 16% of the variance in PSQI Total score at long-term follow-up. In the final model, three predictors were retained, including sex (i.e., female), YGTSS Tic-related Impairment score, and ADHD-RS-IV Total score. Of these, YGTSS Tic-related Impairment score (β = .34) was a statistically significant predictor of PSQI Total score (see Table 3 and Supplementary Table 3). Sex and ADHD-RS-IV Total score were retained in the final model despite p-values >.05 based on parsimony; together, these three variables reflected the simplest model while preserving adjusted R². The final model explained 15% of the variance in PSQI Total score.

In the full model, SCARED-Adult Total score (β = .38) was a significant cross-sectional predictor of PSQI Total at long-term follow-up (see Table 4 and Supplementary Table 4). The full model explained 44% of the variance in PSQI Total score at long-term follow-up. Three predictors were retained in the final reduced model; chronological age, SCARED-Adult Total score (β = .40), ASRS Total score (β = .31)) were positive predictors of PSQI Total score at long-term follow-up (See Table 4 and Supplementary Table 4). The final reduced model explained 42% of the variance in PSQI Total score.