Obsessive-Compulsive Symptoms and Cigarette Smoking: An Initial Cross-Sectional Test of Mechanisms of Co-Occurrence

Abstract

Tobacco addiction and obsessive-compulsive symptoms (OCS; intrusive thoughts or impulses that cause distress and rituals) are both mediated by compulsivity and negative reinforcement. Little evidence exists to guide theory, research, treatment, and population-based prevention of this co-occurrence. We propose a conceptual model of OCS-smoking co-occurrence in which smokers with elevated OCS capitalize on smoking to assuage OCS. This secondary analysis examined associations between OCS levels and self-reported smoking motives that are model-concordant: i) negative reinforcement—smoking for relief; ii) sensorimotor—benefits of behaviors and sensations of the tobacco self-administration ritual; and iii) habitual—smoking without conscious control. A community sample of cigarette smokers enrolled in a cessation trial (N=458; 47.2% female; M_age=36.9; SD=13.6) completed pre-quit self-report measures of OCS levels and smoking motives. Regression models adjusted for sociodemographic covariates and non-OCS psychopathologies indicated that OCS level was positively associated with each model-concordant motive. OCS level was also correlated with tobacco dependence severity and greater withdrawal symptom severity experienced during previous quit attempts. Those with higher OCS report greater motivation to smoke for negative reinforcement, sensorimotor behavioral-ritualistic, habit/automaticity, and stimulation reasons. Further examination of the proposed model of OCS-smoking co-occurrence is warranted to guide theory and intervention for this population.

Introduction

Compulsivity and negative reinforcement are core features of both tobacco addiction and obsessive-compulsive symptoms (OCS; intrusive thoughts or impulses that lead to distress and rituals executed to alleviate such distress; Hall et al., 2015; Pauls, Abramovitch, Rauch, & Geller, 2014). Yet, existing evidence to guide theory, research, treatment, and population-based prevention of the co-occurrence between OCS and tobacco use remains remarkably thin.

Like other psychopathologies that co-occur with smoking (Subramaniam et al., 2009), OCS levels vary along a continuum in the general population (Haslam, Williams, Kyrios, McKay, & Taylor, 2005). While the prevalence of meeting DSM diagnostic criteria for obsessive-compulsive disorder (OCD) in population-based samples is low (12-month prevalence ~1%; Adam, Meinlschmidt, Gloster, & Lieb, 2012; Ruscio, Stein, Chiu, & Kessler, 2010), an appreciable portion of the population experience ‘subclinical’ levels of OCS (~8% to 20%; Adam et al., 2012; Ruscio et al., 2010; Subramaniam et al., 2009; Torres et al., 2006). The OCS construct can be operationalized quantitatively by the number of different types of symptoms experienced over the past week and their relative severity (Foa et al., 2002;).

Research on the association between OCD and smoking status is mixed. Small non-representative studies have failed to find an association between OCD and smoking incidence (Abramovitch, Pizzagalli, Geller, Reuman, & Wilhelm, 2015; Baker-Morissette, Gulliver, Wiegel, & Barlow, 2004), although some early reports pointed to a possible inverse association (Bejerot & Humble, 1999; Himle, Thyer, & Fischer, 1988), suggesting OCD tendencies may protect against smoking. However, stronger evidence of a positive association of OCD or OCS level with the incidence of smoking or tobacco dependence has been found in several large epidemiologic studies (Adam et al., 2012; Grabe et al., 2001; Jaisoorya et al., 2015; Lawrence, Mitrou, & Zubrick, 2009; Subramaniam et al., 2009; Torres et al., 2006).

Regardless of whether OCS is associated with the incidence of smoking, OCS levels could perpetuate and exacerbate tobacco addiction amongst established regular smokers (i.e., notwithstanding OCS as a protective or risk factor for smoking onset, OCS may be associated with smoking maintenance). A recent study found that African American smokers with higher OCS reported greater levels of nicotine dependence, more cigarettes smoked per day, and more severe tobacco withdrawal symptoms after experimentally-induced tobacco abstinence (Bello, Pang, Chasson, Ray, & Leventhal, 2017). Beyond this paper, published evidence or theory on the mechanisms by which OCS could perpetuate tobacco addiction amongst established smokers is absent, leaving scientists and clinicians without a primer for addressing OCS-smoking co-occurrence.

To address this gap, this study presents an initial test of a novel theoretical model of the mechanisms linking OCS and persistent tobacco addiction. The model purports that, in regular smokers, OCS perpetuates smoking due to the negatively reinforcing behavioral-sensorimotor effects that accompany smoking and are uniquely salient to smokers with elevated OCS. The tobacco self-administration ritual (i.e., sequence of lighting the cigarette, hand-to-mouth movements, puffing, inhalation, and exhalation) often acquires negatively reinforcing distress-alleviating properties independent of nicotine (Perkins, Karelitz, Conklin, Sayette, & Giedgowd, 2010). Because individuals with elevated OCS are impelled to carry out compulsive ritualistic and habitual acts as a mechanism to quell distress (Pauls, et al., 2014), they may find the effects of tobacco self-administration sequence and its associated sensorimotor stimuli especially reinforcing. Furthermore, they may be more compelled to execute the smoking ritual habitually and automatically in response to distress.

As an initial test of this model, this secondary analysis leverages data from daily cigarette smokers enrolled in a smoking cessation trial (Zvolensky et al., 2018; Zvolensky et al., 2017) that included a pre-treatment measure of OCS and the Reasons for Smoking Scale (RFS), which assesses six qualitatively-distinct motives for smoking (Ikard, Green, & Horn, 1969). We hypothesized that smokers with higher OCS levels report the following ‘model-concordant’ RFS motives as more salient: (1) negative reinforcement (i.e., smoking to alleviate negative affect), (2) sensorimotor (i.e., benefits of behaviors and sensations associated with the smoking self-administration ritual), and (3) habitual (i.e., tendency to smoke automatically without conscious control). We also expected null associations between OCS and two other smoking motives that are hypothesized not to be relevant to smokers with elevated OCS. Those two motives include positive reinforcement (desire for pleasure) and addiction (tendency to consciously experience strong cigarette cravings). To add to the sparse evidence base on OCS-smoking co-occurrence, a secondary aim of this study was to report correlations between OCS and clinically-relevant indicators of tobacco addiction (i.e., cigarettes smoked per day, nicotine dependence severity, and withdrawal or other problems experienced during prior quit attempts), as well as the associations between smoking indicators and scores on the various dimensions of OCS, given its heterogeneity (McKay et al., 2004).

Methods

2.1. Participants

Participants were 458 treatment-seeking smokers enrolled in a smoking cessation trial involving the provision of behavioral cessation counseling and nicotine-replacement therapy (Zvolensky et al., 2018; Zvolensky et al., 2017). The current report is based on secondary analyses of baseline (pre-treatment) data. Eligible participants were required to be ≥18 years old, report smoking an average of ≥8 cigarettes daily for ≥1 year, and endorse motivation to quit smoking of ≥5 on a 10-point scale. Exclusion criteria included current receipt of smoking cessation treatment, current active suicidal intention, and history of psychotic disorders.

2.2. Procedure

Treatment-seeking adult daily smokers in Burlington, VT and Tallahassee, FL USA were recruited from the community with flyers, newspaper ads, and radio announcements to participate in a dual-site randomized controlled clinical trial examining the efficacy of a standard smoking cessation treatment versus an integrated treatment for smoking and anxiety (Schmidt, Raines, Allan, & Zvolensky, 2016). Participants provided written informed consent, were administered the SCID-I/NP, and then completed self-report measures, including those that are the focus of the present report. As part of the clinical trial, participants received $12.50 for the baseline assessment, from which data for the current study were drawn. Though, by completing all assessments during the course of the clinical trial, including its seven follow-up periods, participants could earn up to $142.50. The study protocol was approved by the Institutional Review Board at each study site.

2.3. Measures

2.3.1. Primary Regressor: Obsessive-Compulsive Inventory—Revised (OCI-R)

The OCI-R (Foa et al., 2002) instructs respondents to indicate the degree of distress for 18 OCS items on a 5-point scale ranging from 0 (Not at all) to 4 (Extremely). The OCI-R consists of a Total (severity) Scale based on the sum all 18 items, as well as six 3-item subscales, reflecting different symptom types: Washing (e.g., “I wash my hands more often and longer than necessary.”), Checking (e.g., “I repeatedly check doors, windows, drawers, etc.”), Ordering (e.g., “I need things to be arranged in a particular way.”), Obsessing (e.g., “I find it difficult to control my own thoughts.”), Hoarding (e.g., “I collect things I don’t need.”), and Neutralizing (e.g., “I feel I have to repeat certain numbers.”). We used the continuous OCI-R scores that indicate OCS level for the primary analysis and report the proportion of the sample who surpassed the OCI-R Total Scale clinical cutoff score (21+) for descriptive purposes. The OCI-R has been shown to have excellent test-retest reliability and strong convergent validity (Foa et al., 2002).

2.3.2. Primary Study Outcome: Smoking Motives.

The Reasons for Smoking Scale (RFS; Ikard et al., 1969) assesses six conceptually-distinct motives for smoking by separate subscale composite scores: Habitual (4 items; e.g., “I smoke cigarettes automatically without even being aware of it”), Addictive (5 items; e.g., “I get a real gnawing hunger for a cigarette when I haven’t smoked in a while”), Pleasurable Relaxation/Positive Reinforcement (2 items; e.g., “Smoking cigarettes is pleasant and relaxing”), Reduction of Negative Affect/Negative Reinforcement (6 items; “When I feel uncomfortable or upset about something, I light up a cigarette”), Stimulation (3 items; e.g., “I smoke cigarettes to stimulate me, to perk myself up”), and Sensorimotor (3 items; e.g., “Handling a cigarette is part of the enjoyment of smoking it”). Items are rated on a 5-point Likert scale ranging from 1 (never) to 5 (always), and the RFS has demonstrated good convergent and factorial validity in previous research (Tate & Stanton, 1990).

2.3.3. Smoking Characteristics

The Fagerström Test for Cigarette Dependence (FTCD; Heatherton, Kozlowski, Frecker, & Fagerstrom, 1991) is a well-validated 6-item self-report measure of tobacco cigarette dependence severity that assesses heavy and compulsive smoking behavior. Sum scores range from 0 to 10, and higher scores indicate greater severity of tobacco dependence.

The Smoking History Questionnaire (SHQ; Brown, Lejuez, Kahler, & Strong, 2002) includes survey items assessing number of cigarettes smoked per day during the past week (Cig/day: 1 item) and a composite score based on the sum of 5-point ratings of depression, difficulty concentrating, irritability, increased eating, decreased heart rate, and other problems or tobacco withdrawal symptoms experienced during previous quit attempts (Withdrawal; 17 items).

2.3.4. Covariates and Descriptive Measures

Ten planned covariates were included in each model to determine the specificity of the OCS-smoking associations based on prior evidence of their respective relations with either OCS or smoking-related characteristics: age (Fidler, Jarvis, Mindell, & West, 2008; Ruscio et al., 2010), sex (Fidler et al., 2008; Ruscio et al., 2010), race (Bello et al., 2017), relationship status (Fidler, et al., 2008), non-nicotine substance use disorder (Ruscio et al., 2010), and domains of distress, such as depression, panic, social anxiety, and traumatic intrusions (Jaisoorya et al., 2017; Ruscio et al., 2010).

An author-constructed questionnaire included standard items assessing sex, race, age, and education level.

The Structured Clinical Interview for DSM-IV-TR Axis I Disorders Non-Patient Edition (SCID-I/NP; First, Spitzer, Gibbon, & Williams, 2007) was administered by trained research assistants or doctoral-level staff in order to assess for past year Axis I psychopathology. Interviews were audiotaped, and the reliability of a random selection of 12.5% of interviews was checked for accuracy; no cases of diagnostic disagreement were noted. We report OCD diagnosis for descriptive purposes and use non-nicotine substance use disorder as a covariate.

The Inventory for Depression and Anxiety Symptoms (IDAS; D. Watson et al., 2007) is a 64-item measure that assesses specific symptom dimensions related to DSM-IV-based major depression- and anxiety-related disorders. The IDAS includes subscales assessing Panic, Social Anxiety, Traumatic Intrusions, and General Depression. Items ask participants to rate the extent to which they experienced symptoms (e.g., “I felt inadequate.”) during the “past two weeks” on a 5-point Likert scale (1=Not at all to 5=Extremely). Mean rating per item was reported for each subscale. The IDAS has demonstrated strong internal consistency and convergent and discriminant validity in previous research (Watson et al., 2008; Watson et al., 2007).

2.4. Data Analysis

Descriptive analyses and tests of the secondary aim of examining relations with smoking characteristics (besides motives) involved reporting correlations of OCI-R Total Scale with smoking characteristics and other psychopathology measures. For primary analyses, linear regression models were tested in which OCI-R Total Scale and covariates (e.g., depression, age, sex) were included as simultaneous regressors, with separate models tested for each smoking motive. For comparative purposes, associations between psychopathology measure covariates and smoking motives are also reported. In a supplemental analysis to determine the extent of OCS-smoking associations across different OCS symptom types, the same regression modeling strategy was utilized with separate models for each combination of OCI-R subscale and smoking-related outcome after adjusting for covariates. Results are reported as standardized regression coefficients with standard error of estimates (βs±SE). Raw p-value thresholds are presented for all tests (<.05, <.01, <.001), and the Benjamini-Hochberg method (Benjamini & Hochberg, 1995) was applied to determine whether each association of the OCI-R total scale was significant after correction for multiple testing of 10 outcomes to keep study-wise type-I error at .05.

Results

3.1. Descriptive Analysis and Association of OCS with Smoking Characteristics

Descriptive statistics for sample characteristics and Cronbach’s alphas for study measures are shown in Table 1. The sample was on average 36.9 (SD=13.6) years old, 47.2% female, and mainly (91.5%) White race, with an average age of initiating regular smoking of 17.4 (SD=3.9) and years having been a regular smoker of 18.7 (SD=13.3) years. The M and SD scores on the IDAS and OCI-R in our sample are comparable to other community samples and indicate low levels, on average, but substantial inter-individual variability in psychopathology (Foa et al., 2002; Watson, et al., 2008; Watson, et al., 2007). Fifty-six (12.2%) of participants surpassed recommended cutoffs of indicating clinically-significant OCS. Current DSM-IV OCD diagnoses based on SCID-I/NP interviews were rare (2.2%), but consistent with epidemiological estimates of past-year OCD prevalence of 1.2% (Ruscio et al., 2010), and split evenly between men (n=5) and women (n=5). The point-biserial correlation between OCI-R Total Scale and OCD diagnostic status in the current sample was moderate (r_pb=.31, p<.001), paralleling previous work and owing to the wide degree of variance in ‘subclinical’ OCS in the overwhelming majority of the population who do not meet OCD criteria (Foa et al., 2002).

Table 1.

Sample Descriptive Statistics and Correlations of OCI-R with Smoking Characteristics and Other Variables

	M (SD) or %		Cronbach’s α		Correlation with OCI-R Scale (r)a
Sociodemographics
Age (years)	36.9 (13.6)		-		−.06
Female gender	47.2%		-		−.08
Race			-		.07a
White	91.5%
Black	8.5%
Attended College	71.4%		-		−.14*
Married or in Live-in Relationship	34.5%		-		−.01
Obsessive compulsive symptoms
OCI-R Total Scale	10.1 (9.7)		.91		-
OCI-R Washing	0.9 (1.7)		.76		.68**
OCI-R Checking	1.4 (2.2)		.85		.79**
OCI-R Ordering	2.6 (2.6)		.89		.77**
OCI-R Obsessing	1.8 (2.3)		.83		.74**
OCI-R Hoarding	2.6 (2.7)		.83		.69**
OCI-R Neutralizing	0.8 (1.8)		.79		.71**
Other psychopathologies
IDAS General Depression	40.8 (13.4)		.93		.56**
IDAS Panic	11.1 (4.3)		.87		.61**
IDAS Social Anxiety	8.0 (3.8)		.88		.51**
IDAS Traumatic Intrusions	5.9 (2.8)		.85		.53**
Non-nicotine substance use disorder	14.4%		-		.07
Smoking characteristics and motives
FTCD	5.2 (2.3)		.65		.12*
SHQ Cigarettes smoked per day	16.9 (10.2)		N/A		.03
SHQ Withdrawal in Past Quit attempts	34.9 (11.6)		.91		.34**
RFS Habitual	9.3 (2.9)		.71		.26**
RFS Addictive	16.5 (3.9)		.79		.24**
RFS Positive Reinforcement	7.5 (1.6)		.83		.04
RFS Negative Reinforcement	20.7 (4.8)		.88		.33**
RFS Stimulation	8.0 (2.8)		.83		.25**
RFS Sensorimotor	7.5 (3.0)		.78		.23**

Note. N = 458; OCI-R = Obsessive-Compulsive Inventory Revised; IDAS = Inventory of Depression and Anxiety Symptoms; FTCD = Fagerström Test of Cigarette Dependence; SHQ = Smoking History Questionnaire; RFS = Reasons for Smoking scale. N/A= Not applicable.

^aPearson correlation for all estimates except for associations for categorical variables, which are point-biserial correlation values (coded as follows: gender [female = 1, male = 0], race [White = 1, Black = 0], education level [attended college = 1, no college= 0], relationship status [married or in live-in relationship = 1, other relationship status = 0], non-nicotine substance use disorder [yes = 1, no = 0]).

^*p < .01,

^**p < .001

Pearson correlation coefficients between OCI-R measures and other psychopathologies are shown in Table 2. OCI-R total score and subscale measures were moderately to strongly correlated with each other and with the IDAS scales. Unadjusted Pearson correlations between OCI-R and each of the study outcomes show statistically significant associations with each smoking motive and characteristic (rs range=.12 to .34, ps<.01), with the exception of Cig/day (r=.03) and the RFS-Pleasurable Relaxation/Positive Reinforcement subscale (r=.04).

Table 2.

Associations of OCI-R and Other Forms of Psychopathology with Smoking Motives

		Standardized Regressor Estimates, β (SE)
				Other Psychopathologies
Outcome Measure		OCI-R Total Scale		IDAS-General Depression		IDAS-Panic		IDAS-Social Anxiety		IDAS-Traumatic Intrusions		Non-nicotine SUD		Overall Model R2
Model-concordant motives
RFS Habitual		.15 (.06)*a		.04 (.07)		.12 (.07)		−.03 (.06)		.06 (.07)		−.05 (.05)		.15***a
RFS Negative Reinforcement		.17 (.05)**a		.30 (.06)***a		−.02 (.06)		.02 (.06)		.03 (.06)		−.05 (.04)		.28***a
RFS Stimulation		.15 (.06)*a		.06 (.07)		.03 (.07)		.06 (.07)		.06 (.07)		−.10 (.05)*		.11***a
RFS Sensorimotor		.18 (.06)**a		−.04 (.07)		.04 (.07)		−.02 (.06)		.10 (.06)		.01 (.04)		.23***a
Other motives
RFS Addictive		.09 (.06)		.10 (.07)		.00 (.07)		.09 (.07)		.11 (.07)		−.02 (.05)		.13***a
RFS Positive Reinforcement		.11 (.06)		−.05 (.07)		−.13 (.07)		.04 (.07)		.05 (.07)		−.01 (.05)		.05*a

Note. N = 458; OCI-R = Obsessive-Compulsive Inventory Revised; IDAS = Inventory of Depression and Anxiety Symptoms; RFS = Reasons for Smoking Scale. SUD = Substance Use Disorder. SE = Standard Error. Standardized regressor estimates from linear regression models that included OCI-R Total Scale, each IDAS subscale, and all other study covariates (i.e., age, sex, race, education level, relationship status, and non-nicotine substance use disorder) as simultaneous regressors. Separate models tested for each outcome.

^*p < .05,

^**p < .01,

^***p < .001

^aStatistically significant after Benjamini-Hochberg correction for multiple testing

3.2. Primary Analysis: Association of Overall OCS Levels and Smoking Motives.

Regression models including the OCI-R scale and the 10 covariates yielded significant omnibus overall model effects for the set of 11 total regressors for each outcome (R²s=.05 to .28; ps<.001). The parameter estimates for OCI-R Total Scale score adjusted for study covariates was significant for regressions involving each model-concordant motive (i.e., negative reinforcement, habitual, sensorimotor, stimulation), indicating a positive relation to higher scores for each motive (see Table 2). The magnitudes of association were small to medium in magnitude (βs=.15 to .18) and significant after type-I error correction. While the unadjusted correlations of the two other smoking motives (addictive and positive reinforcement) with OCI-R were significant (see Table 1), the estimates from regression models adjusting for other psychopathologic and sociodemographic variables were non-significant (βs=.09 to .11). Examination of the parameter estimates for the psychopathology covariate associations with each RFS outcome revealed only one association—IDAS General Depression was significantly associated with RFS Negative Reinforcement scores.

3.3. Supplemental Analyses: Association of OCS subtypes and Smoking Motives

Descriptive results and results from adjusted linear regressions that examined the association of each individual OCI-R symptom subscale with smoking motives in separate models can be found in Supplementary Tables 1 and 2. Most OCS subtypes were associated with some of the model concordant RFS motives. However, these associations were, on average, of smaller magnitude than those involving the OCI-R Total Scale and most consistent for Obsessing and Hoarding symptom subtypes, with partial evidence of associations with Checking and Neutralizing symptom subtypes. The association between Washing symptoms and study outcomes was uniformly null.

Discussion

In this secondary analysis of pre-quit data from a smoking cessation trial, the pattern of associations between OCS levels and smoking motives were consistent with a model of putative behavioral mechanisms linking OCS and persistent tobacco addiction. Associations between OCS and model-concordant smoking motives were robust after controlling for sociodemographic factors and other psychopathologies, suggesting a qualitatively-specific link involving OCS, per se, rather than a non-specific association with generalized psychiatric distress. We also report new evidence on the association of OCS with nicotine dependence severity a tendency toward more severe withdrawal symptoms and other problems experienced in prior quit attempts, which replicates previous results and extends them to a sample seeking smoking cessation treatment (Adam et al., 2012; Bello et al., 2017; Grabe et al., 2001; Jaisoorya et al., 2015; Lawrence et al., 2009; Subramaniam et al., 2009; Torres et al., 2006). Additionally, novel data showing that the associations of OCS with smoking motives were somewhat consistent across OC subtypes are presented here.

Individuals with OCS are impelled to carry out ritualistic, automatized behavior as a mechanism to reduce distress (Pauls et al., 2014). Our OCS-smoking model used this premise to hypothesize that smokers with elevated OCS may find the behavioral-sensorimotor ritual of smoking—the sequence of lighting a cigarette, hand-to-mouth movements, puffing, inhalation, and exhalation—particularly reinforcing. Concordant with this hypothesis, smokers with higher OCS in this study reported a greater tendency to execute the smoking ritual automatically (RFS-Habitual) and value the behavioral sequence and associated stimuli of lighting up and puffing (RFS-Sensorimotor).

One smoking motive assessed in the current study— the propensity to value the attention- and arousal-enhancing effects of nicotine, which is measured by the RFS-Stimulation scale—also associated positively with OCS. Based on this finding, the proposed model of OCS-smoking co-occurrence might benefit from integrating the pharmacological effects of nicotine, which involve increased cognitive control (Hall et al., 2015), which may help mitigate the tendency of smokers with elevated OCS to become absorbed in obsessional cues.

We provide partial evidence of discriminant validity within the OCS-smoking model in that the OCS was not associated with other, non-model concordant, motives, including pleasure-seeking (i.e., positive reinforcement) or craving-related (i.e., addiction) motives. While null results could indicate no association, the optimal method of demonstrating discriminant validity would be to conduct head-to-head comparisons in the association estimates between model-concordant and other motives. We lacked sufficient power to conduct such head-to-head comparisons. Thus, null results do not definitively indicate the absence of an association. Additionally, the unadjusted correlation between OCS with addiction motives was statistically significant (r=.23, p<.001), and the adjusted association was modest in magnitude but not definitively negligible (β=.09, p=.12). Thus, addiction-related motives may warrant further inquiry into the pathogenesis of OCS-smoking co-occurrence.

The unadjusted correlation with positive reinforcement motives was small (r=.04). Conceptualizations of obsessive-compulsive phenomena predominantly focus on negative reinforcement mechanisms—rather than pleasure-seeking behavior—as key underpinnings of the syndrome (Mowrer, 1947), as the affective consequences of obsessions and compulsions in OCS involve distress alleviation but not pleasure enhancement. Given this notion and the negligible association with the positive reinforcement RFS scale, it is likely that any role of affect-mediated reinforcement in OCS-smoking co-occurrence is predominately specific to negative affect and negative reinforcement.

Despite the heterogeneity of OCS phenomena (McKay et al., 2004), the overall pattern of results from the subscale analysis indicate that associations of OCS with model-concordant smoking motives and characteristics were fairly consistent across several OCS symptom components (Supplementary Table 2). Negative reinforcement motive scores were associated with OCI-R Obsessing and Ordering. These subscales reflect autogenous obsessionality in which the predominant distress to the individual stems from a meta-cognitive aversion to the intrusive thought (i.e., in Obsessing) or urge (i.e., in Ordering) itself (Lee & Kwon, 2003). Nicotine-induced cognitive control that facilitates negative-reinforcement mediated smoking by distracting from these aversive meta-cognitive states could be particularly salient for smokers with elevated obsessing or ordering OCS. Findings for some other subscales require further investigation, including the absence of associations between the OCI-R Washing subtype and study outcomes.

Study limitations include the cross-sectional observational design, which preclude temporal or causal inference. It is possible that unmeasured confounds account for the associations presented. For example, environmental factors linked to smoking relapse (e.g., smoking environments; Ngo et al., 2019) could account for the associations presented. Yet, the comprehensive covariate adjustment reduces the likelihood of residual confounding. Reverse causation is also possible, such that experiencing certain reinforcing effects from smoking exacerbate OCS and warrants inquiry. For example, smoking may trigger health-related anxiety in individuals with OCS who experience intrusive thoughts about the adverse effects of smoking, and this could lead to compulsive checking or other OCS-based rituals (Abramowitz, Brigidi, & Foa, 1999).

Future longitudinal and experimental studies on OCS-smoking co-occurrence may address these questions. The nearly exclusive use of self-report measures introduces the possibility that common-method bias may partially account for some of the associations. Furthermore, this study did not measure physical symptoms such as severity of respiratory symptoms, which has been found to be associated with anxiety and depressive symptoms (Ho, Tan, Ho, & Chiu, 2019). Because the sample predominately included White heavy, dependent smokers interested in quitting, current findings may not fully generalize to minority group smokers (e.g., African Americans), and restriction of range in smoking frequency in the sample may have attenuated associations with cig/day or dependence severity if OCS contributes to variation at the lower end of the tobacco addiction severity spectrum.

The study results do not address the small fraction of people who meet diagnostic criteria for OCD. Rather, this study addresses the mechanisms underlying smoking perpetuation associated with variation along the dimensional construct of OCS, which has a wide continuum of functioning in the population. Low OCS levels may be fairly common and involve experiencing obsessions or compulsions only periodically or with very little distress.

In sum, this study provides initial support for a proposed OCS-smoking maintenance model that establishes a framework for understanding this sorely understudied co-occurrence. Future validation of the model would point to intervention considerations for the population of smokers with OCS elevations, such as behavioral and pharmacological interventions that promote alternate methods of distress alleviation or replace or extinguish the sensorimotor-behavioral reinforcement caused by the smoking ritual. For example, promising electronic health interventions for smoking cessation (e.g., Tran et al., 2018) might benefit from including treatment approaches targeting OCS. Additional work may point to novel considerations in understanding the etiology of tobacco addiction more broadly. Ritualistic behavior and negative reinforcement are core features of both addiction and OCS (Hall et al., 2015; Pauls et al., 2014). It is possible that the tendency toward OCS-like behavior, which often manifests early in life ( Zohar, 1999) could represent a marker of vulnerability to addiction. To this end, the current study represents an early step toward studying the implications of OCS and ritualistic behavior in tobacco addiction.