A Cross-Sectional Study of How Harm Avoidance, Incompleteness and Intolerance of Uncertainty Contribute to Obsessive–Compulsive Disorder in University Students

: Research suggests that certain cognitive factors increase the likelihood of developing and maintaining obsessive–compulsive disorder (OCD). Such factors that are often associated with OCD are harm avoidance (HA), incompleteness (INC), and intolerance of uncertainty (IU). The present study aimed to examine the associations of intolerance of uncertainty, incompleteness, and harm avoidance with dimensions of obsessive–compulsive symptoms. Participants were 1128 university students (Mage = 19.42 St.d. = 2.02). Results showed that all subscales of the OCI-R correlated significantly with HA, INC, and IU, confirming the findings of previous studies. HA and INC appear to predict OC symptoms in a significant way. IU appears to correlate and improve the models, although to a lesser degree. The present findings contribute to our better understanding of the relationship between OCD symptoms and underlying cognitive variables and the nature of OCD heterogeneity.


Introduction
Obsessive-compulsive disorder (OCD) is often described as a heterogeneous disorder with a thematic array of intrusions and obsessions [1].Clinicians and researchers suggest that OCD symptoms may be classified into distinctive subtypes: (1) contamination-related obsessions and washing/cleaning rituals, (2) responsibility for harm and checking rituals, (3) a need for symmetry or order and arranging rituals, and (4) unacceptable thoughts (e.g., violence-or sex-related) and covert mental neutralising [2].
Theoretical models were proposed [3][4][5] that focus on cognitive factors, which appear to increase the likelihood of developing and maintaining OCD.Many suggest that obsessions occur when an otherwise unnecessary and unwelcome cognitive intrusion (e.g., the thought of harming a loved one) is misinterpreted catastrophically.For example, e.g., the interference can be misjudged as morally inappropriate or lead to an unintentional consequence.However, the attempt to dismiss the thought or alleviate the anxiety caused by the unwanted thought increases the preoccupation with the intrusive thoughts, which, in turn, does not allow their disconfirmation [6].

Harm Avoidance and Incompleteness
Researchers have recently begun to recognize the influence of underlying motivational factors that drive maladaptive behaviours and may be important in OCD, such as harm avoidance (HA) and incompleteness (INC) [23].
Summerfeldt et al. [24] proposed that harm avoidance and incompleteness are core dimensions that, either singly or in combination, underlie all OC phenomena.They proposed that there is no specific correspondence between the compulsion and the underlying driver.More specifically, individuals may repeatedly check whether they did a kitchen task in order to reassure themselves that they are safe.Nevertheless, they may just as well repeatedly check to ensure it was performed correctly until it feels complete [24][25][26][27].In this context, various compulsions can be understood in relation to these two driving core dimensions.
Harm avoidance (HA) is a heritable tendency to respond strongly to unpleasant stimuli, causing the inhibition of behaviours that may lead to punishment, novelty, or frustration [28].HA is a personality trait that has been described as a vulnerability factor for several anxiety disorders [29] and has been associated with anxiety, worrying, and a desire to prevent potential harm [30].
OCD patients demonstrate higher HA, often associated with increased severity in obsessive-compulsive symptoms [31][32][33][34].Furthermore, people with OCD and their firstdegree relatives showed significantly elevated scores of HA, supporting the idea that HA is a risk factor for familiarity with OCD [35].
On the other hand, incompleteness is "an inner sense of imperfection, connected with the perception that actions or intentions have been incompletely achieved" ( [36] p. 80).Rasmussen and Eisen [37] identified abnormal risk assessment, pathological doubt, and incompleteness as primary components of OCD.Drawing on their work, Coles et al. [38] and Summerfeldt [39] noted that a broad subcategory of patients seem to be bothered by feelings of incompleteness or "Not Just Right Experiences" (NJREs) as a core phenomenon rather than by irrational anxiety.Subsequently, Summerfeldt et al. [30] proposed a twodimensional "core affective-motivational model".Recent findings support the Motivation Model of OCD and the notion that the motivational domains may be related to different levels of beliefs and symptoms [40].Ecker and Gönner [26] proposed that the initiation of rituals prompted by obsessions may hinge on the degree of harm avoidance, while the perpetuation of these rituals is primarily influenced by a sense of incompleteness.The level of harm avoidance could indicate how individuals respond to intrusive thoughts, triggering anxiety and prompting anxiety-alleviating rituals.On the other hand, the levels of incompleteness may determine how easily behavioural sequences and rituals, whether induced by anxiety or other factors, can be terminated once they have begun [41].
Washing, checking, and neutralising compulsions are associated with both HA and INC [26,42,43], though one motivational factor may dominate.Incompleteness is associated with higher symptom severity [44] and a poorer response to therapy [45].Therefore, patients with more severe overall symptomatology are more likely to have prominent INC [25].Coles et al. [46] proposed that the desire for things to feel complete or "just right" plays an essential role in transitioning from initial obsessions and compulsions to full-blown OCD.
INC-driven OCD symptoms may result from a failure of a "stop signal" process, which marks that the behaviour reached its intended end and ought to stop [47,48].Consequently, persistent "error signals" may inappropriately prompt ongoing corrective action [49,50].In recent research with OCD patients, findings suggested that symptoms associated with feelings of incompleteness appear to be related to deficits in executive functioning and problem-solving [51].
A recent study showed that treatment reduced incompleteness and harm avoidance, and changes were positively correlated with changes in OCD severity.Importantly, when accounting for covariance between variables, incompleteness alone was uniquely associated with a change in OCD severity [52].

Intolerance of Uncertainty
Another factor that appears to have a critical role in OCD is the intolerance of uncertainty (IU).IU refers to the tendency to display negative responses in the face of uncertain situations encompassing cognitive, emotional, and behavioural levels [53].Individuals with high IU are inclined to view a new, unpredictable future with apprehension, perceiving uncertainty as a source of threat [16].At the cognitive level, those high in IU tend to interpret ambiguous situations pessimistically, anticipating adverse events and experiencing negative emotions like anxiety and frustration [54].Behaviourally, individuals with elevated IU may either avoid uncertain situations or engage in actions aimed at resolving ambiguity.Research suggests that IU is a transdiagnostic cognitive bias [55] that plays a role in the symptoms of several psychiatric disorders, including both Generalized Anxiety Disorder [56][57][58] and obsessive-compulsive disorder (OCD; [16,59]).A recent meta-analysis by McEvoy et al. [60] concluded that IU is a transdiagnostic process with a strong association with various disorders, including OCD.
People with obsessive-compulsive symptoms seem to be very careful and need more time to sort objects and more often ask for information to be repeated [61].Also, they have more doubts about the correctness of their decisions [62].Difficulties in decisionmaking can arise from beliefs about the need for certainty.The intolerance of uncertainty is characterized by the tendency of the individual to fear a new, unpredictable, and uncertain future and the belief that the feeling of uncertainty is threatening.Intolerance to uncertainty is seen in various forms of psychopathology and OCD [16].
Studies showed that IU predicted OC symptoms above responsibility, control, and threat estimation, even when controlling for depression and anxiety symptoms [59,63].Furthermore, in a study with undergraduate students, IU was most strongly related to OC symptoms even when controlling for health anxiety and sensitivity to anxiety and neuroticism [64].
Furthermore, a study with in vivo exposures showed that IU predicted the urge to check; in a contamination test, it predicted avoidance, the urge to wash, and the duration of washing.In a sorting test, it predicted the urge to tidy and the duration of tidying.It did not predict neutralising [65].Moreover, a recent study showed that IU mediates the association between NJREs and OCD in a clinical sample.Results supported the role of NJREs as motivators of OCD, and those IU beliefs mediate the association between NJREs and OCD probably because of the need to achieve a sense of certainty [66].
The present study aimed to examine the associations of intolerance of uncertainty, incompleteness, and harm avoidance with specific OC symptoms.To our knowledge, few studies incorporate these three factors and their predictability of OC symptoms.

Participants
In total, 1128 participants were included in the study, aged 18 to 35 (Mage = 19.42St.d.= 2.02).They were all university students in Thessaloniki; 81.5% were from the University of Macedonia and 18.5% from the Aristotle University of Thessaloniki.Participants were from various schools across campus, of which 61.5% were first-year students, 28.9% were second-year, 3.6% were third-year, and the rest, 5.9% were seniors in their final years of university.Most participants were female, consisting of 65.3% of the total sample.Participation in the study was voluntary, and participants were given no money or credits.The Ethics Committee of the University of Macedonia approved the research.

Measures
Obsessive Compulsive Inventory-Revised (OCI-R; [67]).OCI-R measures the severity of OCD symptoms.It comprises 18 items divided into six sub-scales: washing, checking, or-dering, hoarding, neutralisation, and obsessions.Participants' scores are rated on a 5-point scale (0 = Not at all, 1 = A little, 2 = Moderately, 3 = A lot, 4 = Extremely).In the current study, we used the Greek version of the OCI-R, which has proven good psychometric properties [68].The internal consistency of the OCI-R was excellent, with Cronbach's Alpha α = 0.86.
Obsessive-Compulsive Trait Core Dimension Questionnaire (OC-TCDQ; [30]).The OC-TCDQ is a 20-item self-report measure assessing HA (10 items) and INC (10 items).Each item is rated from 0 ("Never applies to me") to 4 ("Always applies to me").In the current study, we used the Greek version of the OC-TCDQ.The scale was translated from English to Greek and back to English by two authors (AN and GS).Finally, any differences were discussed until they reached a consensus.The accuracy of these translations was reviewed by the study's third author (MS).In the present study, the OC-TCDQ showed excellent internal consistency with a Cronbach's Alpha of α = 0.92 and yielded a two-factor solution.
Intolerance of Uncertainty Scale-Short Form (IUS-12; [69]).IUS-12 is a 12-item selfreport measure evaluating one's tendency to find uncertainty upsetting and distressing, and each item is scored on a 5-point Likert scale ranging from 1 (not at all characteristic of me) to 5 (entirely characteristic).IUS-12 consists of two factors: (1) Prospective Anxiety (7 items) and (2) Inhibitory Anxiety (5 items).Despite the reported multifactor structures, the IUS is commonly summed up as a total scale score [70].The IUS-12 has proven excellent internal consistency and convergent and discriminant validity with a solid bifactor structure [71].The IUS-12 showed excellent internal consistency in the present study with a Cronbach's alpha of α = 0.87.

Procedure
Participants were all university undergraduates who were recruited from university classes.Participants were informed about the study's aim and purpose in their classrooms.Participation was voluntary, and individuals who agreed to participate were given a hard copy booklet, which included (1) a consent form, (2) a demographic form, (3) OCI-R, (4) IUS-12, and (5) OC-TCDQ.A written consent form was obtained from all participants.The authors of the study digitalized participants' responses.SPSS and Amos performed the data analysis.

Statistical Analysis
For the statistical analysis, the statistical software packages SPSS 21 and Amos 21 were utilized.Firstly, the factory structure of TCDQ was examined.Bentler and Chou [72] propose that a ratio of 5 cases per variable is adequate for normally distributed data.A commonly endorsed guideline suggests having a minimum of ten observations per indicator variable to ensure an acceptable sample size [73].Consequently, considering those mentioned above, and since the observed variables were 20 items, we randomly extracted 25% from the primary data pool with SPSS 21.0, which resulted in 290 participants using Amos to perform Confirmatory Factor Analysis (CFA).CFA was used to test the goodness-of-fit of the hypothesized latent structure of the TCDQ (i.e., two correlated factors, each comprised of ten items).The residuals of the three items loading on each factor were correlated in the model.We examined two-factor solutions using maximum likelihood estimation with the correlation matrix.Model fit was determined using (1) chi-square, (2) Comparative Fit Index, (3) goodness of fit, (4) Standardized Root Mean Square Residual, (5) Root Mean Square Error of Approximation [74].
Descriptive statistics were computed, normality checks were conducted with most variables following a normal distribution, except for OCI_Neutralising and OCI_washing (Kolmogorov-Smirnov < 0.05), and Pearson's correlation analysis was performed to examine relationships between variables.Linear regression modelling was then employed to further explore these relationships.The dependent variables were the OCI-R and subscales, and the independent variables were harm avoidance, incompleteness, and IU.
The Durbin-Watson statistic, nearing 2 (1.992), indicated minimal autocorrelation in the residuals, supporting the assumption of independence.However, collinearity diagnostics highlighted potential multicollinearity concerns among harm avoidance, incompleteness, and intolerance of uncertainty.Visual inspection of the residual plots, including the residuals vs. fitted values plot and scale-location plot, did not provide evidence against homoscedasticity.

Factor Analysis
The goodness of fit statistics are reported in Table 1.The original two-factor model (Table 1) had an adequate fit.Most standardized factor loadings were above 0.50, except for item 20 (loading was 0.48) ranging from 0.48 to 0.79.The two factors were significantly correlated: r = 0.792, p < 0.001.The model fit was a reasonably good fit, with most fit indices being permissible; therefore, it was deemed appropriate for the present research.The fit index of GFI is sensitive to larger sample sizes and may present a downward bias [75].

Descriptive Statistics and Correlations
Table 2 shows participants' means and standard deviations and Cronbach's alpha in all scales and their respective subscales.Table 3 shows correlations for all OCI-R subscales with HA, INC, and IUS-12.Table 4 shows correlations among the independent variables (HA, INC, and IUS).

Regression Analysis
A stepwise regression was used to examine which factors can predict OCD symptom severity as a total (Table 5) and each symptom dimension separately (Table 6).

Discussion
The present study examined the associations and predictive value of intolerance of uncertainty, harm avoidance, and incompleteness with OC symptoms in a Greek nonclinical young adult sample.
The proposed factor structure of the OC-TCDQ was empirically supported in this study, as previous findings suggest [30].The two-factor structure model yielded acceptable results.The CFI and GFI indexes were borderline unacceptable; however, the literature shows that these bounds are not absolute and should be interpreted as indications that the model fits the data well [76].With the other indices showing good fit, we concluded that the measure had adequate factor structure for this study without further modifications.
The internal consistency was high for all scales (OCI-R, OC-TCDQ, and IU) and good for all the subscales.Only three (hoarding, neutralising, obsessing) subscales of the OCI-R reported lower Cronbach's alpha, which is consistent with previous findings that also showed lower values that were still above 0.50 [77][78][79][80].
All subscales of the OCI-R correlated significantly with HA, INC, and IU, confirming the well-known relationship between OCD symptoms and relevant belief domains [26,65].In the stepwise regression analysis, HA and INC predict OC symptoms significantly.IU appears to improve the models, however, to a lesser degree.From all symptom dimensions, ordering shows the strongest INC association, while washing was substantially associated but to a somewhat weaker extent.On the other hand, obsessing showed no significant association with INC.
More specifically, hoarding appeared to have a strong association with HA and far less with INC and IU.Previous studies suggest that there is a non-significant association between INC and hoarding [26,39,81].Although our results support this finding, it is appropriate to recognize that recent studies suggest that the hoarding dimension is not central to the symptom dimension network of OCD [82].This is in line with the current classification of hoarding disorder as a mental disorder distinct from OCD [83], suggesting that further studies on hoarding are needed, which will be more specific to the conceptualization of hoarding, like whether HA is associated with other dimensions relating to hoarding, i.e., the avoidance of grief [84].
Checking was associated with both INC and HA.IU appeared to have a smaller effect than the other factors.These results are consistent with the claim that checking is associated with HA and INC [26,42,43].However, as other studies suggest, the IU did not appear as strong as HA and INC in predicting checking [65].However, many researchers hypothesized that HA likely motivated checking [25,27,85].The present results are consistent with Ecker and Gonner [26], who found that HA and INC both predicted checking; their main hypothesis to explain this finding was that while HA may trigger the initiation of checking rituals, INC may be responsible for the subsequent failure to terminate them.The hypothesis was also supported by Summerfeldt [39], who further differentiated harm-avoidant checking from checking as a compulsion in general, suggesting that the latter could also be explained by INC [81].A consensus seems to be that checking may be motivated by either HA or INC in a particular individual [23].Even though IU was found to predict compulsions relating to checking in multiple studies, it is interesting that when it is in a model with HA and INC, it does not show unique associations above them.This finding may be explained by the idea that IU has been found to increase threat expectancy and biased threat appraisal [86].Therefore, IU may increase HA tendencies.
INC is the strongest predictor of OCI-R Ordering.This pattern of results is consistent with the previous literature that indicates not-just-right phenomena have a prominent role in ordering/symmetry [87][88][89].Moreover, Radomsky and Rachman [90] emphasize the rarity of harm-related cognitions in their subjects with elevated ordering scores; this may explain that HA showed a negative association with ordering even though no one reported such a strong relationship.Recent research found incompleteness predicted ordering even after controlling for HA [91].The results imply that incompleteness may be more important to some types of OCD symptoms.Finally, IU did not seem to have a statistically significant impact on ordering.
HA and INC have significant positive influences on neutralising, with HA having a slightly more robust influence.Cognitive symptoms of OCD usually show a more robust HA association [26].However, recent findings [89] show that INC has a strong association even after controlling for HA.IU also has a positive influence but is less pronounced than HA and INC.However, this model explains approximately 17.9% of the variance.This pattern of results is consistent with the previous literature showing that neutralising does not have a unique association with either HA or IN [26].
INC, and to a lesser degree, HA, strongly predicted washing, while IU was not associated with washing.Washing appears to have mixed HA and INC motivation.Washing may include fears of harm by becoming contaminated [26].Ólafsson et al. [92] suggest that not-just-right experiences and levels of incompleteness partially mediate the relationship between disgust and contamination fear.As was mentioned in checking, HA may motivate the initiation of washing rituals, while INC perpetuates the rituals.A study showed that induced INC/NJREs correlated with compulsion severity and were associated with ordering, washing, and hoarding symptoms [93].
Obsessing was uniquely associated with HA, far less with IU, but not with INC.The present results are consistent with Schreck et al.'s [94] findings that showed HA had unique associations, predominantly with cognitive symptoms such as doubting, obsessing, and neutralising symptoms.One interpretation of these findings is that the cognitive-behavioural model of OCD emphasizes the role of HA beliefs like inflated personal responsibility and the overestimation of threat [95] as prominent in the development of OCD.
Research suggests that there is a large proportion of OCD patients, around 40%, that do not try to avoid harm, but they feel discomfort when they try to discontinue their rituals [91,96].Two studies with large clinical OCD samples identify "low beliefs" subgroups of OCD sufferers ([97] 56% of the sample, [98] 51% of the sample) "who do not report elevated personal responsibility, threat estimation, perfectionism, intolerance of uncertainty, or over-importance and over-control of thought" ([97] p. 1357).Our findings highlight the possibility that these patients may be associated with feelings of incompleteness or "not just right experiences".This idea is further supported by the fact that some patients with OCD show resistance to conventional CBT therapy, which consists of a strong cognitive component closer to harm-avoidant symptoms.In a recent study, the initial findings from the comparison between harm avoidance (HA) and incompleteness (INC) generally point toward the conclusion that current implementations of cognitive-behavioural therapy (CBT) for OCD may not effectively address incompleteness symptoms [99].
On the other hand, Foa et al. [45] report a trend toward poorer behaviour therapy outcomes for OCD symptoms without a harm-avoidant component (only 45% symptom reduction vs. 69% for harm-avoidant OCD).Hence, assessing motivational factors that may have important implications for therapeutic interventions is essential.Furthermore, Sperling [100] showed that IU before CBT therapy was an indicator of the effectiveness of therapy, with participants with higher IU showing less improvement throughout therapy.Even though IU did not emerge as a prominent factor in the present research, it is vital to understand its role in OCD.In terms of future research, it would be useful to extend the current findings by examining in a prospective study which factor may predict OCD symptom severity and the relationship that these cognitive factors have among them.
Several key limitations should be noted.Firstly, the participants were all university students, so the results might not necessarily be representative.Other factors that may also affect university students, such as substances, were not considered in this study.Secondly, conclusions concerning the OCD specificity of INC and HA are premature because of the lack of clinical and non-clinical control groups.Finally, the cross-sectional nature of our data precludes causal inferences.Despite these limitations, the present study enhanced our understanding of the relationship between OCD symptoms, HA, INC, and IU.We hope that the current research will stimulate further investigation of this important area.Our results align with Summerfeldt's [39,81] model, which proposed that INC and HA may be motivational "core dimensions" contributing to a better understanding of OCD heterogeneity.Unfortunately, IU did not emerge as a robust predicting factor.

Table 2 .
Means and standard deviations for age, all scales, and subscales.

Table 5 .
Stepwise regression with OCI-R total as DV and incompleteness, harm avoidance, and intolerance of uncertainty as independent variables.

Table 6 .
Linear regression analysis with dependent variable OCI-R subscales and independent variable incompleteness, harm avoidance, IU.