Cross-Disorder Comparison of Sensory Over-Responsivity in Chronic Tic Disorders and Obsessive-Compulsive Disorder

Abstract

Background:

Sensory over-responsivity (SOR) refers to excessively intense and/or prolonged behavioral responses to environmental stimuli typically perceived as non-aversive. SOR is prevalent in several neurodevelopmental disorders, including chronic tic disorders (CTDs) and obsessive-compulsive disorder (OCD). Few studies have examined the extent and clinical correlates of SOR across disorders, limiting insights into the phenomenon’s transdiagnostic clinical and biological relevance. Such cross-disorder comparisons are of particular interest for CTDs and OCD given their frequent co-occurrence.

Objective:

We sought to compare the magnitude of SOR between adults with CTD and adults with OCD and to identify the clinical factors most strongly associated with SOR across these disorders.

Methods:

We enrolled 207 age- and sex-matched participants across four diagnostic categories: CTD without OCD (designated “CTD/OCD−”; n=37), CTD with OCD (“CTD/OCD+”; n=32), OCD without tic disorder (“OCD”; n=69), and healthy controls (n=69). Participants completed a self-report battery of rating scales assessing SOR (Sensory Gating Inventory, SGI), obsessive-compulsive symptoms (Dimensional Obsessive-Compulsive Scale, DOCS), inattention and hyperactivity (Adult ADHD Self-Report Screening Scale for DSM-5, ASRS-5), anxiety (Generalized Anxiety Disorder-7), and depression (Patient Health Questionnaire-9). CTD participants were also administered the Yale Global Tic Severity Scale (YGTSS). To examine between-group differences in SOR, we compared SGI score across all groups and between pairs of groups. To examine the relationship of SOR with other clinical factors, we performed multivariable linear regression.

Results:

CTD/OCD−, CTD/OCD+, and OCD participants were 86.7%, 87.6%, and 89.5%, respectively, more likely to have higher SGI total scores than healthy controls. SGI total score did not differ between CTD/OCD−, CTD/OCD+, and OCD groups. In the regression model of log-transformed SGI total score, OCD diagnosis, DOCS score, and ASRS-5 score each contributed significantly to model goodness-of-fit, whereas CTD diagnosis and YGTSS total tic score did not.

Conclusion:

SOR is prevalent in adults with CTD and in adults with OCD but does not significantly differ in magnitude between these disorders. Across CTD, OCD, and healthy control adult populations, SOR is independently associated with both obsessive-compulsive and ADHD symptoms, suggesting a transdiagnostic relationship between these sensory and psychiatric manifestations. Future cross-disorder, longitudinal, and translational research is needed to clarify the role and prognostic import of SOR in CTDs, OCD, and other neurodevelopmental disorders.

1. INTRODUCTION

Chronic tic disorders (CTDs), such as Tourette syndrome (TS), are defined by the presence of tics, but sensory symptoms are also characteristic.[1,2] Sensory phenomena in CTDs can be divided into premonitory urges and sensory over-responsivity (SOR). A premonitory urge is an uncomfortable bodily sensation building in the seconds before a tic and, typically, resolving with execution of the tic.[1,3] Such urges are experienced by 90% of adults and adolescents with a CTD.[3,4] The severity of premonitory urges correlates with the severity of tics in most,[5–8] but not all,[9] studies. In contrast to premonitory urges, which are discrete, transient phenomena temporally linked with tics, SOR is a temporally pervasive phenomenon not time-locked to tics. SOR is characterized by excessively intense and/or prolonged behavioral responses to environmental stimuli that are not perceived as aversive by neurotypical individuals.[10,11] The presence of SOR indicates an imbalance between sensitization (i.e., progressively enhanced response to a stimulus with repeated exposure) and habituation (i.e., progressively weakened response to a stimulus with repeated exposure).[12] Notably, other labels have been applied to the SOR phenomenon, including sensory hypersensitivity [2,13,14] and sensory intolerance.[1] Examples of environmental stimuli that may distress an individual with SOR include noise from electrical appliances; clothing tags or seams rubbing against the skin; and bright indoor lighting.[14,15] As many as 50% of children with CTDs [16,17] and 80% of adults with CTDs endorse some degree of SOR.[13,14]

SOR is evident in a number of neurodevelopmental disorders besides CTDs, including obsessive-compulsive disorder (OCD),[18–21] attention deficit hyperactivity disorder (ADHD),[22,23] autism spectrum disorder (ASD),[24–26] and schizophrenia.[27,28] It remains uncertain whether SOR is merely an associated feature arising from neural substrates shared across these disorders or whether SOR plays a causal role in the emergence and/or trajectory of the disorders’ core phenotypes. While still preliminary, mounting evidence implicates the latter. SOR emerges early in childhood,[12,29,30] and sensory processing, more broadly, is known to be critical for motor,[31] social,[31–34] and emotional development.[31,32,35] SOR has been linked with impairments in social interaction [32] and family life,[36] as well as with internalizing symptoms.[37–39] Longitudinal cohort studies reveal that SOR predicts future anxiety symptoms in children from the general population [29] and in children with ASD.[40] No similar SOR longitudinal research has been conducted in any adult populations. In cross-sectional studies, extent of SOR has been shown to correlate positively with severity of core phenotypic symptoms in several neurodevelopmental disorder populations, including children with OCD,[20] adults with ADHD,[41] and adults with ASD.[24] In adults with CTDs, SOR does not correlate with tic severity, but it does correlate with severity of obsessive-compulsive and ADHD symptoms,[13] which are integral dimensions of the CTD phenotype. In TS, lifetime rates of OCD and ADHD are 50% and 54%, respectively,[42] and many with TS and other CTDs who do not satisfy formal diagnostic criteria for OCD and ADHD display sub-clinical symptoms of these disorders.[43] Thus, SOR is common among neurodevelopmental disorders, often parallels core phenotype severity, and appears to hold prognostic value for early childhood anxiety symptoms.

Despite its cross-cutting prevalence in neurodevelopmental disorders, studies of SOR to-date have generally focused on single disorders (CTD,[13,14] OCD,[19,20] ADHD,[41] or ASD [24–26]), with few notable exceptions.[44,45] In one cross-disorder investigation, Ludlow et al found that sensory sensitivity in children with TS (n=12) and in children with ASD (n=12) differed from healthy controls (n=12), but the study was underpowered to identify significant differences between the TS and ASD groups.[44] A study of children 3–14 years old identified no differences in SOR between participants with ASD (n=77) and those with ADHD (n=78), who were matched by sex and age.[45] Despite the relatively large body of SOR research in OCD, ADHD, and ASD populations, we were unable to locate other studies examining the phenomenon across these disorders. Comparing findings across studies is hampered by heterogeneity in study methodology. In particular, a variety of rating scales are used to assess SOR. Commonly employed self-report measures of SOR include the Adolescent/Adult Sensory Profile,[41] SenSOR Inventory,[36] and Sensory Gating Inventory.[13,46] The most widely used proxy-report measures are the Short Sensory Profile [17,20] and the SenSOR Inventory (which is adaptable for self- and proxy-report). These scales vary in their psychometric properties (e.g., factor structure and item response options) and in the degree to which they assess each sensory modality, a notable discrepancy given SOR may not be uniform across the senses.[47–49] Even researchers administering the same scale may diverge in their analysis and interpretation of the score, contingent upon their treatment of SOR as a dimensional [20,29,41,50] or categorical [29,36,49] phenomenon. Thus, given the abundant evidence that SOR is widespread across neurodevelopmental disorders, cross-disorder studies using the same assessment procedures promise to deepen insights into the transdiagnostic clinical relevance and neuroscientific implications of SOR. Such efforts align with the National Institute of Mental Health (NIMH) Research Domain Criteria Initiative (RDoC), with its emphasis on studying common dimensions of mental health disorders.[51]

Previously, we demonstrated that SOR is more frequently reported by adults with CTDs (n=34) relative to healthy controls (n=34) and that SOR in adults with CTDs is closely associated with obsessive-compulsive symptoms.[13] For the current study we sought to contrast the extent of SOR between adults with CTDs and adults with OCD and to identify the clinical factors most strongly associated with SOR across these disorders.

We elected to study adults, as opposed to children, with CTD and/or OCD because adults with persistent symptoms of these disorders represent a more severe phenotype.[52–55] Such a symptomatically enriched sample facilitates evaluation of the relationship between clinical variables. Moreover, there is growing recognition of the need to understand neurodevelopmental disorders across the entire lifespan, not solely in childhood.[52,54–60] To achieve our study objectives, we recruited four groups of adults: adults with both CTD and OCD, adults with CTD without OCD, adults with OCD without tic disorder, and healthy controls. Participants completed self-report rating scales indexing SOR and psychiatric symptoms common in CTDs and OCD. We hypothesized that those with CTD and/or OCD would have significantly elevated SOR scores relative to healthy controls, in keeping with the abundant literature evidencing SOR in neurodevelopmental disorder populations. We further hypothesized that individuals with both CTD and OCD would exhibit the greatest SOR among the four participant groups, extrapolating from pediatric CTD studies that show psychiatric comorbidity compounds magnitude of SOR.[16,17] Lastly, we hypothesized that obsessive-compulsive symptoms would be the clinical factor most strongly associated with SOR across all participants, given evidence of such an association in studies that focused on individual disorders (CTDs [13]; OCD [20]) and in studies of the general population.[50]

2. MATERIALS AND METHODS

2.1. Participants

From April 2019 through June 2021, we prospectively recruited adults (>18 years of age) with CTDs from Vanderbilt University Medical Center (VUMC) Tourette Syndrome Clinic and from VUMC research registries. CTD, as a diagnostic category, comprises the following Diagnostic and Statistical Manual of Mental Disorders, 5^th edition (DSM-5) diagnoses: TS, chronic (persistent) motor tic disorder, and chronic (persistent) vocal tic disorder. These three disorders exist on a single clinical continuum.[61] Tic disorders were diagnosed by a movement disorders neurologist using DSM-5 criteria. CTD participants were divided into those with self-reported OCD (CTD/OCD+) and those without self-reported OCD (CTD/OCD−).

Between April 2020 and June 2021, we recruited two additional participant groups: 1) healthy adults and 2) adults with an established OCD diagnosis but no history of tic disorder (denoted “OCD” for the remainder of the manuscript). Healthy control and OCD participants were recruited via the web-based registry ResearchMatch,[62] VUMC research registries, and the International OCD Foundation website (www.iocdf.org). Healthy control and OCD participants completed all study assessments online (as discussed below in Section 2.2) and did not undergo diagnostic psychiatric interviews. All participants were explicitly asked to report prior diagnoses of tic disorders, OCD, ADHD, ASD, anxiety, and/or depression. Healthy control and OCD participants were one-to-one matched on sex and age (± 10 years) to CTD participants. Only participants who completed at least 50% of study measures were included in the matching process. Healthy control and OCD participants with a self-reported diagnosis of tic disorder were excluded from matching, as were healthy control participants with self-reported OCD, ADHD, or ASD. Healthy controls with anxiety and/or depression were included in the matching process. Only data from matched participants were included in the analyses.

English fluency was requisite for study enrollment. All participants gave informed consent electronically and were compensated monetarily for their time. The VUMC Institutional Review Board approved this study. Initial findings from the first enrolled CTD (n=34) and healthy control (n=34) participants were published previously.[13]

2.2. Measures

Table 1 lists the rating scales used in this study. A movement disorders neurologist (D.I.) experienced with the Yale Global Tic Severity Scale (YGTSS) administered the scale to all CTD participants. The YGTSS is the gold-standard rating scale for tic severity. After completing the YGTSS, CTD/OCD− and CTD/OCD+ participants were emailed a unique hyperlink to the battery of self-report rating scales and asked to complete the scales at their earliest convenience, in a single sitting. Eligible healthy control and OCD recruits were provided participant-specific hyperlinks to the same self-report rating scale battery. Self-report rating scales were administered via Research Electronic Data Capture (REDCap), a HIPAA-compliant online platform and database.[63,64] Estimated time to complete the entire online battery was 20–35 minutes.

Table 1.

Study rating scales

Scale	Number of Items	Score Range†
Sensory Gating Inventory (SGI)[65]	36	0–216
Dimensional Obsessive-Compulsive Scale (DOCS)[70]	20	0–80
Adult ADHD Self-Report Screening Scale for DSM-5 (ASRS-5)[105]	6	0–24
Generalized Anxiety Disorder-7 (GAD-7)[109]	7	0–21
Patient Health Questionnaire-9 (PHQ-9)[110]	9	0–27
Yale Global Tic Severity Scale Total Tic Score (TTS)[111]*	10	0–50

^†For each scale, higher score indicates more of the clinical phenomenon being measured.

^*Administered only to CTD participants

No widely accepted gold standard measure of SOR exists. For this study, we selected the Sensory Gating Inventory (SGI) to quantify SOR given its rigorous psychometric development and validation.[65] The SGI is a self-report questionnaire consisting of 36 statements about sensory perception. Respondents are asked to rate their agreement with each statement on a 6-point Likert scale from “never true” to “always true.” Higher total score signifies greater SOR. No established score cutoff demarcates normal from abnormal. The SGI was developed (n>1,000 participants) and validated (n>800 participants) in large samples of healthy adults, with the final scale demonstrating good test-retest reliability, convergent validity (with measures including the Highly Sensitive Person Scale), and discriminant validity.[65] The SGI questions predominantly focus on the auditory and visual modalities. We previously showed in an adult CTD sample that SGI correlates closely (Spearman rank correlation r_s = −0.73) with the Sensory Perception Quotient,[66] a measure of sensory hypersensitivity that includes items for tactile, olfactory, gustatory, visual, and auditory modalities.[13] As determined by confirmatory factor analysis, the SGI has a four-factor structure, with domains for Perceptual Modulation; Distractibility; Over-Inclusion and Hyperawareness; and Fatigue and Stress Vulnerability.[65] Notably, SGI scores correspond with neurophysiologic markers of sensory gating impairment (see Section 4, “Discussion”) in schizophrenia [67] and ADHD.[68,69]

We quantified severity of obsessive-compulsive and ADHD symptoms with validated self-report measures (Table 1). A score cutoff of 20 on the Dimensional Obsessive-Compulsive Scale (DOCS) yields 70% sensitivity and 70% specificity in distinguishing OCD from other anxiety disorders.[70] A score cutoff of 13 on the Adult ADHD Self-Report Screening Scale for DSM-5 (ASRS-5) yields 81% sensitivity and 70% specificity in detecting ADHD in clinical populations.[71] As adults with CTDs frequently suffer from anxiety and/or depression, which thus represent potential confounds, we also assessed these psychiatric symptoms with validated self-report measures (Table 1).

2.3. Statistics

Medians and interquartile ranges are provided as measures of central tendency for continuous variables. Missing responses for individual scale items were imputed from mean, non-missing responses from the entire participant pool. To assess internal consistency reliability for the SGI, coefficient omega was calculated.

To examine between-group differences in SOR, we compared SGI total score and SGI domain scores across all groups (Kruskal-Wallis rank sum test). Significance thresholds for omnibus tests (10 statistical tests) were Bonferroni-adjusted to correct for multiple comparisons, resulting in significance thresholds of p < 0.005. Significance thresholds for post hoc pairwise comparisons (Wilcoxon-rank sum test) were set at p < 0.05. We did not correct for multiple comparisons in post hoc testing (for pairwise comparisons or for the regression modeling analysis discussed below) given the exploratory nature of the study.[72] A power analysis for post hoc pairwise comparisons is provided in the Supplemental Material. Magnitude of the Wilcoxon rank-sum test statistic served as a non-parametric measure of effect size.[73]

To quantify the association between the SGI, its four domains, and the psychiatric symptom scales (DOCS, ASRS-5, GAD-7, PHQ-9) across all participants, we computed Spearman’s rank correlations (r_s) for scale scores, applying Bonferroni correction to the significance threshold to account for multiple comparisons (26 statistical tests, yielding a significance threshold of p < 0.0019). For CTD participants, we also calculated Spearman’s rank correlations of YGTSS total tic score (TTS) with the other scale scores (9 statistical tests; p < 0.0056).

To examine the relationship of SOR with clinical factors across participants, we used multivariable linear regression. Given increasing dispersion of the dependent variable (SGI total score) with increasing values of several independent variables, we log-transformed SGI total score. Independent variables for the base model (designated “Model 1”) were selected based on evidence of an association with SOR from the published literature (DOCS, ASRS-5, CTD diagnosis, OCD diagnosis) or plausibility of an association with SOR (age, sex, TTS). Study sample size was insufficient to permit modeling with interaction terms and medication effects. We fit the model using data from all participants, rather than testing models for each group separately, to quantify the strength of the symptom interrelationships across disorders. For the regression analysis, healthy control and OCD participants were assigned a TTS of zero since each of these participants denied any history of tic disorder. Three nested models were constructed from the base model by setting select coefficients to zero, to determine the impact of removing CTD diagnosis and TTS (Model 2), OCD diagnosis and obsessive-compulsive symptoms (Model 3), and ADHD symptoms (Model 4). A fifth model was constructed by adding GAD-7 and PHQ-9 as independent variables to the base model (Model 5). Regression diagnostics performed for each model included plotting histograms of residuals against a normal curve to visually examine for deviations from normality; plotting residuals against independent variables to visually examine for heteroscedasticity; calculating Breusch-Pagan test statistic to quantify heteroscedasticity; and calculating variance inflation factor (VIF) for each independent variable to assess for multicollinearity. Significant multicollinearity was prespecified as VIF > 5.[74] Model goodness-of-fit was indexed by adjusted R² and Akaike information criteria (AIC). Smaller AIC values indicate less information loss.[75] The significance threshold for model goodness-of-fit was Bonferroni-adjusted to account for multiple comparisons (5 models with ln(SGI) as the dependent variable, resulting in a significance threshold of p-value < 0.01). Likelihood ratios were used to compare goodness-of-fit between the base model and the other regression models. For each model, we performed post-hoc t-tests examining the association of the independent variables with log-transformed SGI total score; pre-specified significance threshold was set at p < 0.05.

Statistical analyses were conducted in STATA 15.0.

3. RESULTS

Sixty-nine individuals with CTD, 86 individuals with self-reported OCD (and no history of tic disorder), and 90 healthy controls completed more than half of the surveys. A subset of participants from the OCD and healthy control pools were selected to permit one-to-one age- and sex-matching to the CTD participants. Our final cohort comprised 69 individuals with CTDs (37 with CTD/OCD−; 32 with CTD/OCD+), 69 individuals with OCD, and 69 healthy controls. Demographic and clinical data by group are provided in Table 2 for all matched participants. Of the CTD participants, 65 met criteria for TS, 4 for chronic motor tic disorder, and 0 for chronic vocal tic disorder. Prevalence of self-reported comorbidities in the CTD group was as follows: 46% with OCD, 30% with ADHD, 72% with anxiety, and 59% with depression. CTD participants completed surveys a median of 3 days following YGTSS administration (interquartile range 0 – 14 days). For the final cohort, data were 100% complete for all rating scales, except for DOCS (1 missing item from 1 respondent), GAD-7 (7 missing items across 6 respondents), and PHQ-9 (3 missing items across 2 respondents). Pairwise contrasts of continuous variables are provided in the Supplemental Material. Internal consistency reliability for the SGI scale was excellent for the entire sample of 207 matched participants (coefficient omega = 0.97).

Table 2.

Demographic and clinical characteristics by group

	Healthy Controls (n=69)	CTD/OCD- (n=37)	CTD/OCD+ (n=32)	OCD (n=69)	Omnibus Test for Continuous Variables†
Sex (M:F)	39:30	22:15	17:15	39:30	-
Age	30 (24–43)	31 (22–41)	28 (21–42.5)	29 (25–42)	χ2(3)=0.85
Race
American					-
Indian/Alaska
Native					-
Asian	0	0	0	0
Native
Hawaiian or	8	0	1	0	-
Other	0	0	0	0	-
Pacific					-
Islander	7	0	0	2
Black or	51	37	30	66
African American	2	0	1	1
White	1	0	0	0
More Than
One Race
Unknown / Not
Reported Ethnicity
Hispanic or Latino	7	0	0	0	-
Not Hispanic or Latino	60	36	31	4	-
Unknown / Not	2	1	1	65
Reported
Self-Reported
Diagnoses
ADHD	0	11	10	9	-
OCD	0	0	32	69	-
Anxiety	5	23	27	47	-
Depression	5	21	20	45	-
ASD	0	0	2	1
				12
ASRS-5	6 (5–8)	14 (11–16)	14.5 (11–17)	12 (10–14)	χ2(3)=100.7**
DOCS†	9 (5–13)	13 (10–21)	28.5 (18.5–34)	30 (20–39)	χ2(3)=101.8**
GAD-7	3 (1–5)	10 (5–15)	12 (6.5–15)	12 (7–15)	χ2(3)=81.2**
PHQ-9	3 (1–5)	8 (5–15)	10 (5–15)	9 (6–13)	χ2(3)=64.3**
TTS	-	23 (15–28)	28.5 (18–34)	-	χ2(1)=2.43

^†Kruskal-Wallis rank sum test with ties

^*p < 0.005

^**p < 0.001

^***p < 0.0001

Correlation matrices are provided in the Supplemental Material. Of the SGI domains, obsessive-compulsive symptom severity correlated most strongly with the Perceptual Modulation domain (r_s = 0.57, p<0.0001), and ADHD symptom severity correlated most strongly with the Distractibility domain (r_s = 0.71, p<0.0001). There were moderate or greater (r_s > 0.40) correlations between scores for SGI total, each SGI domain, DOCS, ASRS-5, GAD-7, and PHQ-9. TTS in the CTD groups correlated weakly with GAD-7 and the SGI Over-inclusion domain, but correlations between TTS and other scale scores did not reach statistical significance after correction for multiple comparisons.

When comparing across all groups (omnibus statistical tests), significant differences were evident for each continuous variable. For pairwise group comparisons, participants with CTD and/or OCD scored significantly higher on OCD, ADHD, anxiety, and depression scales relative to controls (Supplemental Material). The OCD group had significantly lower scores on ASRS-5 compared to the CTD/OCD− and CTD/OCD+ groups (CTD/OCD− vs OCD: z = 2.28, p < 0.05; CTD/OCD+ vs OCD: z = 2.43, p < 0.05). And the CTD/OCD+ and OCD groups scored significantly higher on the DOCS compared to the CTD/OCD− group (CTD/OCD− vs CTD/OCD+: z = −3.88, p < 0.001; CTD/OCD− vs OCD: z = −5.21, p < 0.0001). SGI total score and SGI domain scores were significantly lower for healthy controls compared to the other three groups (Table 3). CTD/OCD−, CTD/OCD+, and OCD participants were 86.7%, 87.6%, and 89.5%, respectively, more likely to have higher SGI total scores than healthy controls. SGI total score did not differ between CTD/OCD−, CTD/OCD+, and OCD groups. For pairwise group comparisons of SGI domain scores, the OCD group had higher significantly scores on the Perceptual Modulation domain than the CTD/OCD− group (CTD/OCD− vs OCD: z = −2.16, p < 0.05). CTD/OCD−, CTD/OCD+, and OCD groups did not differ on any other SGI domain scores.

Table 3.

Sensory Gating Inventory (SGI) total score and domain scores by group

	Healthy Controls	CTD/OCD−	CTD/OCD+	OCD	Omnibus Test & Pairwise Contrasts†
SGI Total	68 (59–81)	112 (88–132)	123 (100–148.5)	125 (105 146)	χ2(3)=82.2** HC vs CTD/OCD−: z=−6.21*** HC vs CTD/OCD+: z=−6.06*** HC vs OCD: z=−8.00*** CTD/OCD− vs CTD/OCD+: z=−1.34 CTD/OCD− vs OCD: z=−1.64 CTD/OCD+ vs OCD: z=−0.047
SGI Perceptual Modulation Domain	21 (18–28)	39 (25–51)	47 (33–58)	46 (35–61)	χ2(3)=81.4** HC vs CTD/OCD−: z=−8.19*** HC vs CTD/OCD+: z=−5.89*** HC vs OCD: z=−8.19*** CTD/OCD− vs CTD/OCD+: z=−1.58 CTD/OCD− vs OCD: z=−2.16§ CTD/OCD+ vs OCD: z=−0.37
SGI Distractibility Domain	19 (13–24)	32 (26–38)	32.5 (23–41.5)	32 (25–37)	χ2(3)=62.7** HC vs CTD/OCD−: z= -5.63*** HC vs CTD/OCD+: z=−5.06*** HC vs OCD: z=−7.12*** CTD/OCD− vs CTD/OCD+: z=−0.46 CTD/OCD− vs OCD: z=0.27 CTD/OCD+ vs OCD: z=−0.69
SGI Over-inclusion Domain	13 (10–19)	21 (18–28)	27 (20.5–35)	26 (19–32)	χ2(3)=56.8** HC vs CTD/OCD−: z=−4.93*** HC vs CTD/OCD+: z=−5.60*** HC vs OCD: z=−6.42*** CTD/OCD− vs CTD/OCD+: z=−1.48 CTD/OCD− vs OCD: z=−1.18 CTD/OCD+ vs OCD: z=0.66
SGI Fatigue & Stress Domain	12 (9–15)	19 (14–22)	19.5 (12.5–23)	20 (16–24)	χ2(3)=45.9* HC vs CTD/OCD−: z=−4.65*** HC vs CTD/OCD+: z=−3.85*** HC vs OCD: z=−6.36*** CTD/OCD− vs CTD/OCD+: z=0.030 CTD/OCD− vs OCD: z=−0.78 CTD/OCD+ vs OCD: z=−0.67

^†Omnibus test statistic provided at top of each cell; pairwise contrast statistics provided below omnibus test

^§< 0.05

^*p < 0.005

^**p < 0.001

^***p < 0.0001

Summary results of the regression analysis are provided in Table 4. Select component plots of log-transformed SGI total score versus continuous independent variables are displayed in Figure 1; fitted, base model regression lines by diagnostic group are superimposed. The base regression model accounted for 50% of the variance in log-transformed SGI total score. Independent variables in each model did not exceed the pre-specified collinearity threshold (VIF < 5) except for TTS (VIF ranging from 5.71 – 5.83) and CTD diagnosis (VIF ranging from 5.61 – 5.80). Substantial collinearity between these two variables precluded unambiguous interpretation of their individual model coefficients. Among the independent variables in the base model, the following were significantly associated with log-transformed SGI total score: OCD diagnosis (β = 0.14), DOCS score (β = 0.0066), and ASRS-5 score (β = 0.036). Notably, in Model 5, these same three variables were also significantly associated with log-transformed SGI total score, but PHQ-9 and GAD-7 scores were not.

Table 4.

Regression analysis summary results

Model Number	Model independent variables^	Breusch-Pagan test†	Model variables significantly associated with ln(SGI)¶	F-test Adj R2	AIC	LR‡
1 (base)	sex, age, CTD diagnosis, OCD diagnosis, TTS, DOCS, ASRS-5	χ2(1) = 1.36 p = 0.24	OCD diagnosis: t = 2.7, p = 0.01 DOCS: t = 3.6, p < 0.001 ASRS-5: t = 6.3, p < 0.001	F(7,199) = 30.9 p < 0.0001 adj R2 = 0.50	70.6	-
2	sex, age, OCD diagnosis, DOCS, ASRS-5	χ2(1) = 0.77 p = 0.38	OCD diagnosis: t = 2.6, p = 0.01 DOCS: t = 3.5, p < 0.001 ASRS-5: t = 8.1, p < 0.001	F(5,201) = 43.1 p < 0.0001 adj R2 = 0.51	67.8	χ2(2) = 1.3 p = 0.53
3	sex, age, CTD diagnosis, TTS, ASRS-5	χ2(1) = 0.03 p = 0.87	ASRS-5: t = 10.5, p < 0.001	F(5, 201) = 29.8 p < 0.0001 adj R2 = 0.41	104.0	χ2(2) = 37.4 p < 0.0001
4	sex, age, CTD diagnosis, OCD diagnosis, TTS, DOCS	χ2(1) = 5.37 p = 0.02	OCD diagnosis: t = 3.7, p < 0.001 DOCS: t = 5.5, p < 0.001	F(6,200) = 24.6 p < 0.0001 adj R2 = 0.41	106.2	χ2(2) = 37.6 p < 0.0001
5	sex, age, CTD diagnosis, OCD diagnosis, TTS, DOCS, ASRS-5, GAD-7, PHQ-9	χ2(1) = 1.89 p = 0.17	OCD diagnosis: t = 2.7, p = 0.007 DOCS: t = 2.0, p = 0.049 ASRS-5: t = 5.1, p < 0.001	F(9,197) = 24.6 p < 0.0001 adj R2 = 0.51	70.8	χ2(2) = 3.8 p = 0.15

^†Null hypothesis of the Breusch-Pagan test is that the variance of the model residuals is constant; p<0.05 indicates significant heteroskedasticity.

^‡Referent for LR is Model 1

^¶Variables not displayed in this column were not significantly associated with ln(SGI) for the given model. Β coefficients are not reported.

^{^}VIF > 5 for TTS and CTD diagnosis in all models where these variables were included

Figure 1.

Base model regression lines, by group

Component regression lines, by group, were adjusted for other base model covariates, using median values for continuous variables and mean value for sex.

Bottom left panel (ln(SGI) vs YGTSS Total Tic Score) depicts only CTD participants.

The likelihood ratio between the base model (Model 1) and the model excluding CTD diagnosis and TTS (Model 2) did not reach statistical significance. The adjusted R² and AIC values for these two models were similar. Taken together, this indicates that the base model and Model 2 fit the data equally well, which suggests that, collectively, CTD diagnosis and TTS did not significantly contribute to model goodness-of-fit. Similarly, the likelihood ratio between the base model (Model 1) and the model including PHQ-9 and GAD-7 scores (Model 5) did not reach statistical significance, and the adjusted R² and AIC values for these models were also similar, suggesting that, collectively, PHQ-9 and GAD-7 scores did not significantly contribute to model goodness-of-fit. In contrast, OCD variables (OCD diagnosis, DOCS score) and ASRS-5 score significantly improved model goodness-of-fit, as evidenced by significant likelihood ratios when comparing models with and without those variables, lower adjusted R² in models without those variables, and higher AIC values in models without those variables. The adjusted R² values were identical for the model without OCD variables (Model 3) and the model without ASRS-5 score (Model 4).

4. DISCUSSION

Results from this prospective study assessing SOR in CTD, OCD, and healthy participants revealed two novel findings. First, contrary to our hypothesis, the extent of SOR did not significantly differ between CTD/OCD−, CTD/OCD+, and OCD adult populations. Second, SOR was independently associated with both obsessive-compulsive symptoms and ADHD symptoms across healthy, CTD, and OCD adult populations. Findings expand upon results from prior studies of SOR in adulthood, which have focused either on single disorders (OCD [19,21]; ADHD [41]) or on a narrow set of symptoms (e.g., obsessive-compulsive symptoms in general population [50]). Findings and their current and future implications are discussed below.

Previous studies show SOR is evident in children [16,17] and adults with CTD [13–15] and in children [20] and adults with OCD.[19,21] Prevalence and magnitude of SOR in these studies, however, vary considerably, in large part due to between-study differences in defining and assessing SOR, in selecting participants (treatment-seeking samples [15,17,20] versus non-treatment-seeking samples [19] versus unreported in study methods [14,21]), and in choosing a control group (study-specific control group [14] versus normative sample from scale validation [17,19,20]). Most studies in CTDs and OCD to-date have categorically defined SOR based on score cutoffs derived from normative samples. For example, the Short Sensory Profile, a widely used proxy-report measure, classifies children into “typical performance,” “probable difference,” and “definite difference” in seven domains, three of which are devoted to heightened sensitivity to touch, taste/smell, and sight/sound.[48] In a sample of 75 children with TS administered the Short Sensory Profile, 58% percent exhibited definite SOR to tactile stimuli, 32% to taste/smell stimuli, and 32% to visual/auditory stimuli.[17] Rates of domain-specific, definite SOR were lower in a sample of 80 children with OCD, also assessed with the Short Sensory Profile: 32.5% displayed SOR to tactile stimuli, 20.5% to taste/smell stimuli, and 20.3% to visual/auditory stimuli.[20] The reported prevalence of SOR in adults with CTDs is higher than that for children with CTDs or OCD, though the aforementioned methodologic differences between studies preclude definitive comparison between adult and pediatric populations. In early seminal work, Cohen et al conducted semi-structured, qualitative interviews on sensory experiences in a mixed sample of 20 adults and children with TS (mean age 20.4 years), noting that 70% of subjects acknowledged some degree of SOR.[15] More recently, Belluscio et al, using an adaptation of the Adult Sensory Profile, identified SOR to visual, auditory, tactile, and olfactory stimuli in over 50% of their sample of 19 TS adults.[14] In adult OCD samples, SOR is evident, but studies to-date have been hindered by multiple methodologic concerns (e.g., inclusion of individuals without diagnosed OCD in the OCD group [19]) or have been limited in scope and data-reporting (e.g., reporting solely statistical relationship between variables without providing SOR raw or scaled scores [21]).

In contrast to several of the above studies,[14,17,20] we adopted a dimensional rather than categorical approach to SOR, given that SOR exists on a continuum. A dimensional approach to SOR permits comparison between individuals who would, under a categorical approach, be classified into the same severity stratum.[76] Consequently, for hypothesis testing, a dimensional approach yields greater statistical power than a categorical approach.[76] The SGI scale does not have established cutoffs for normal and abnormal scores, but over 85% of CTD/OCD−, CTD/OCD+, and OCD participants scored higher on this measure than sex- and age-matched healthy controls. The high percentage of CTD and OCD participants with SGI total scores in excess of control scores further validates SOR as an integral feature of both disorders.

Contrary to our hypothesis that participants with both CTD and OCD would exhibit the greatest SOR, SGI total score did not differ between the CTD/OCD+, CTD/OCD−, and OCD groups. Prior studies in pediatric populations demonstrate that SOR is more severe in CTD [16,17] and OCD [20] patients with psychiatric comorbidities, but no previous such investigations have been undertaken in adults with CTD or OCD. Discrepancies between prior pediatric studies and our current adult study may stem from multiple factors: inherent differences between child and adult CTD phenotypes, distinct assessment methods for SOR and psychiatric symptoms in children and adults (e.g., caregiver-report versus self-report), and/or heterogeneity in participant samples (e.g., recruitment from clinical or non-clinical populations). The OCD group did have higher scores on the SGI Perceptual Modulation domain than the CTD/OCD− group, but patient groups did not significantly differ on any other SGI domain scores. The Perceptual Modulation domain quantifies subjective experience of stimulus intensity,[65] and, thus, study findings suggest that adults with OCD without tics experience this facet of SOR to a greater extent than adults with CTD without OCD. The CTD/OCD+ group also had higher Perceptual Modulation domain scores than the CTD/OCD− group, but that between-group difference did not reach statistical significance (p = 0.11). Ultimately, larger sample sizes on the order of 150–200 participants per group will likely be required to detect potential differences in SOR between adult CTD/OCD−, CTD/OCD+, and OCD groups and to clarify between-group differences in SOR domains (see Supplemental Material for power analysis).

In the regression analysis, we demonstrated that SOR is independently associated with both obsessive-compulsive symptoms and ADHD symptoms across all participants, even after controlling for comorbid psychiatric symptoms. Previously, SOR severity has been observed to correlate with obsessive-compulsive symptom severity in adults with OCD,[21] children with OCD,[20] and adults with CTD,[13] as well as in neurotypical adults.[50,77] However, the majority of these studies did not account for comorbid ADHD symptomatology,[20,21,50,77] despite frequent co-occurrence of OCD and ADHD. Pooled estimates from studies of individuals with OCD without tics indicate the prevalence of comorbid ADHD is 12%, though ADHD symptoms are present in more than 25%.[78] ADHD symptomatology thus represents an unmeasured confound in much of the extant SOR research in OCD. This is significant since extent of SOR has been shown to correlate with severity of ADHD symptoms in several non-OCD populations, including adults with ADHD [41] and adults [79] and children [47] from community samples. Current study findings underscore the importance of assessing both obsessive-compulsive and ADHD symptoms in studies of SOR. Notably, tic severity was not associated with SOR in the regression or the correlation analyses, a finding that aligns with the results of two previous, smaller studies.[15,46] Thus, SOR corresponds to burden of obsessive-compulsive and ADHD symptoms across CTD, OCD, and healthy adult populations. Longitudinal research is needed to determine whether these symptom relationships evolve over the course of development.

Prior pediatric research suggests SOR is a clinical marker for future childhood psychopathology, particularly anxiety.[29,30] Putatively, heightened awareness of and responsivity to sensory stimuli perceived as threatening in early development contributes to chronically maladaptive behavioral and emotional responses.[30,36] In a cohort of 917 community children, degree of SOR at 2–5 years of age predicted severity of anxiety symptoms at 6 years of age.[29] In a sample of 64 children (aged 4–17 years) with anxiety-related disorders, 59% of parents indicated their child first exhibited SOR before age 5.[30] In contrast to studies showing a relationship between SOR and future psychiatric symptoms, a study of 1,046 community children found that caregiver-reported SOR severity at age 8 years did not predict presence of obsessive-compulsive symptoms at age 13.[80] Importantly, due to skewed symptom distributions, the investigators treated obsessive-compulsive symptoms as a dichotomous variable, stratifying participants into those with three or more OCD symptoms and those with two or fewer OCD symptoms. Diagnostic cutoffs based on obsessive-compulsive symptom counts pose significant risk of misclassification bias.[70] Additionally, as OCD has a bimodal age-of-onset, at age 11 years (76% of individuals) or 23 years (24% of individuals),[81] a number of participants in the study who exhibited no or minimal obsessive-compulsive symptoms at age 13 may have developed them later in adolescence. The investigators did observe that more severe SOR at each of the two time points increased the odds of obsessive-compulsive symptoms at that same time point. Given our study results confirming the high prevalence of SOR in adult CTDs and OCD, longitudinal research involving adults is warranted to elucidate the long-term prognostic import of SOR and its potential causal role in psychiatric symptom development and/or exacerbation.

The high prevalence of SOR in adult CTD and OCD populations should also motivate research into its functional and quality of life impact. Anecdotal evidence suggests SOR influences daily activities in these populations, but more rigorous investigations are lacking.[14,15] In adults with ASD, where sensory processing has been more extensively studied, SOR is implicated in anxiety,[82,83] loneliness,[83] and sleep dysfunction,[84] prompting development of adult-based interventions to directly mitigate these symptoms [85] and recommendations to design sensory-considerate work and domestic environments.[24,86] If SOR is confirmed to influence functioning and quality of life in adults with CTD and/or OCD, then it represents an untapped target for novel interventions in many individuals with these disorders.

SOR may also yield insights into shared mechanisms underpinning CTDs and OCD. The precise neural substrates of SOR in CTD and OCD are uncertain, but a common pathophysiologic mechanism is implicated: impaired sensory gating. Sensory gating is the preconscious filtering of redundant sensory stimuli in order to direct cognitive resources to the most novel, and presumably the most salient, environmental input.[65,87] Sensory gating occurs at multiple levels of stimulus processing, involving both bottom-up [88,89] and top-down mechanisms.[89–92] At the anatomic level, prefrontal cortex and primary sensory cortices are key nodes in the sensory gating network.[93–97] The prefrontal cortex regulates attention to sensory stimuli through inhibitory effects on primary sensory cortices.[89,91,94] As evidenced by multiple studies employing different experimental techniques, sensory gating is abnormal in CTDs [1,14,98,99] and OCD.[100–102] In TS, impaired sensory gating has been associated with gamma-amino-butyric acid (GABA) deficiency in primary sensorimotor cortex [103] and with blood oxygen-level dependent (BOLD) hypoactivity in somatosensory processing regions.[96] In OCD, sensory gating impairment correlates with symptom severity, and, furthermore, is partially corrected by deep brain stimulation of the nucleus accumbens, which is posited to influence prefrontal cortex activity via modulation of ventral pallidal output.[104] Additional research is needed to bridge the translational gap between the physiological phenomenon of sensory gating impairment and the clinical phenomenon of SOR in CTDs and OCD. Such efforts promise to shed light on cross-disorder pathophysiological mechanisms.

Our study has several notable limitations. First, results interpretation hinges on the ability of the SGI to accurately quantify SOR. Evidence does strongly support the scale’s construct validity (see Section 2.2). In our sample, the SGI exhibited excellent internal consistency reliability, and our results generally aligned with the extant SOR literature. Second, participant mental health diagnoses were not confirmed with a standardized diagnostic psychiatric interview. Alignment of study results with existing literature from these populations suggests this limitation did not generate extreme bias. Third, we assessed psychiatric symptoms with self-report measures rather than gold-standard, clinician-rated instruments. Nonetheless, the DOCS,[70] ASRS-5,[105] GAD-7,[106] and PHQ-9 [107] are psychometrically sound measures, with good reliability and validity. Fourth, our study population consisted predominantly of non-Hispanic white adults, potentially limiting generalizability of our results to other ethnic and racial groups. A single study has identified possible differences in SOR prevalence by race, with slightly higher rates in minority populations,[108] but subsequent investigations have not replicated this finding.[29,80] Lastly, CTD participants were primarily recruited through a tertiary care clinic, likely resulting in selection of more severely affected individuals and potentially impacting generalizability of study findings to non-clinical CTD populations.

5. CONCLUSIONS

SOR is prevalent in adults with CTD and in adults with OCD but does not significantly differ in magnitude between these disorders. Across CTDs, OCD, and healthy adult populations, SOR is independently associated with both obsessive-compulsive and ADHD symptoms, suggesting a transdiagnostic relationship between these sensory and psychiatric manifestations in adulthood. Future cross-disorder, longitudinal, and translational research is needed to clarify the role and prognostic import of SOR in CTDs, OCD, and other neurodevelopmental disorders.

• Sensory over-responsivity (SOR) is prevalent in chronic tic disorders (CTDs) and obsessive-compulsive disorder (OCD)

• Magnitude of SOR does not significantly differ between CTD and OCD adult samples

• SOR is associated with both obsessive-compulsive and attention deficit hyperactivity disorder (ADHD) symptoms

ABBREVIATIONS:

CTDs

chronic tic disorders

TS

Tourette syndrome

OCD

obsessive-compulsive disorder

ADHD

attention deficit hyperactivity disorder

SOR

sensory over-responsivity

SGI

Sensory Gating Inventory