Ecological Momentary Assessment of Smoking Behaviors in Native and Converted Intermittent Smokers

Abstract

Background and Objectives

About 22% of adult smokers in the U.S. are intermittent cigarette smokers (ITS). ITS can be further classified as native ITS who never smoked daily and converted ITS who formerly smoked daily but reduced to intermittent smoking. Ecologic momentary assessment (EMA) was conducted to determine the behaviors and experiences that are associated with the decision to smoke.

Methods

The study included 24 native ITS and 36 converted ITS (N=60) from the Pennsylvania Adult Smoking Study. A baseline questionnaire, daily log, and an EMA smoking log that assessed emotions, activities, and smoking urges was filled out with each cigarette for one week to capture 574 smoking sessions.

Results

Both groups had very low levels of cigarette dependence. Both groups were more tempted to smoke in positive or negative situations than situations associated with habituation. EMA showed that the most common emotional state during smoking sessions was positive (47%), followed by negative (32%), neutral (16%), and mixed (5%) emotions. Smokers were more likely to smoke during activities of leisure (48%) than during performative duties (29%), social (16%) or interactive occasions (7%). Converted ITS were more likely to smoke alone compared to native ITS (p<.001).

Discussion and Conclusions

ITS report minimal levels of dependence when captured on traditional scales of nicotine dependence, yet experience loss of autonomy and difficulty quitting. The majority of the ITS reported positive emotions and leisure activities while smoking, and smoked during the evening.

1. INTRODUCTION

Cigarette consumption remains the leading preventable cause of morbidity and mortality with nearly 6 million deaths caused by tobacco use worldwide every year.¹ While daily smoking has significantly decreased in the United States, intermittent cigarette use increased from 19.2% to 23.2% between 2005 and 2014.² Despite smoking less frequently, intermittent smokers (ITS) remain at an increased risk of tobacco-related diseases.³ Social or intermittent smoking was recognized and promoted by the tobacco industry starting from the 1970’s.⁴ At that time, intermittent smoking was considered a transient behavior that occurred among youth smokers who would later transition to daily smoking. Over the past few decades, the number of adult smokers who smoke irregularly has increased. ITS appear to avoid the clinical symptoms of withdrawal typically observed with repeated cigarette use.^5,6 Cotinine levels of ITS are about one-third that of daily smokers.⁷ Consequently, it is important to understand and prevent the relative contributions of nicotinic and non-nicotinic factors to smoking maintenance in ITS.

ITS are a diverse group with heterogeneous smoking profiles.⁸ However, distinct phenotypes have been uncovered including “converted” ITS,⁹ who are former daily smokers, many of which converted to intermittent smoking as a transitionary phase to quitting. They are more likely to have used pharmacological aids for cessation, have higher odds of quitting, and show higher motivation towards quitting than other ITS.^10,11 This group is characterized by low nicotine dependence scores, but slightly higher than other ITS,¹² and smoke more cigarettes per day compared to never (native) daily smokers.⁹ “Native” ITS, those who never smoked daily, are less likely to associate themselves as being a smoker ⁴ or initiate total cessation,¹¹ as their consumption is often very low with more days between smoking sessions compared to converted ITS.¹² Importantly, these definitions of ITS are not congruent with past definitions of light smokers such as ‘chippers’, which were defined as those who smoke five or fewer cigarettes per day. The two groups are light smokers, but the distinction lies within the daily use of cigarettes with chippers versus non-daily use in ITS.

Ecologic momentary assessment (EMA) is a methodology that is increasingly being used to study drug dependency including factors that predict smoking cessation.¹³ EMA assesses real-time longitudinal data in the participant’s natural environment and has shown that most ITS are not social smokers, those who smoke primarily while drinking or socializing on weekends.¹⁴ We applied EMA methods to determine how the role of temptations, urges and environmental factors are associated with ITS’ decision to smoke cigarettes, and whether converted ITS have higher levels of these factors than native ITS.

2. METHODS

2.1. Participant

The current study was conducted as part of the Pennsylvania Adult Smoking Study (PASS) ¹⁵ under a separate protocol. Participants were recruited by word of mouth, internet postings, and flyers in central Pennsylvania from 2014 to 2015. Eligibility included smokers aged 25–60, who smoked at least 1 cigarette out of 4–24 days in the past month and had this pattern for the past 6 months. Additionally, anyone who self-reported consuming other forms of tobacco more than 50% of the time, were currently pregnant, or trying to quit (with or without smoking cessation medication or actively reducing amount smoked) were excluded from participation. Fifty percent of subjects were recruited from Craigslist, 18% from the University on-hold phone message and website, 16% from flyers, 14% from word of mouth and 2% from other sources.

2.2. Procedures

All recruitment and procedures were approved by the Pennsylvania State College of Medicine Institutional Review Board. Participants provided written consent and completed a seven-day study protocol. At the initial visit, participants completed a series of self-reported baseline questionnaires. Over the following seven days, participants completed a smoking log, and answered several open-ended and ordinal scale questions during their smoking sessions, and at the end of each day. Participants also collected saliva samples for genetic and nicotine metabolite analysis and used a smoking topography device (data not presented here). At the conclusion of the study, all study materials were collected and participants were compensated for their time. Study data were collected and managed using REDCap (Research Electronic Data Capture) electronic data capture tools hosted at the Penn State Milton S. Hershey Medical Center and College of Medicine.¹⁶

2.3. Measures

2.3.1. Baseline Questionnaire assessments

Participants answered questions on socio-demographics, smoking-related behaviors, medical history, and stress measures. If participants indicated they were ever a daily smoker, they were classified as converted ITS. If participants indicated they were never a daily smoker, they were classified as native ITS. Multiple measures for nicotine dependence were used, including past quit attempts, Fagerström Test for Cigarette/Nicotine Dependence (FTND),¹⁷ Hooked on Nicotine Checklist (HONC),¹⁸ and the 10-item Penn State Cigarette Dependence Index (PSCDI),¹⁹ a scale developed to assess dependency in e-cig and cigarette smokers who exhibit lower thresholds of nicotine dependency. The 20-item Situational Temptation Scale for Smoking,²⁰ which measures how tempted respondents are to smoke in a variety of situations, was asked using a 5-point Likert scale (1=Not At All Tempted to 5 = Extremely Tempted). Examples of situations include smoking over coffee, at a bar, when angry, with a friend, when depressed, or anxious etc. The questions were collapsed into three subscales including negative affect, social/positive, and habit/craving, as well as an overall temptation score.

2.3.2. Ecological Momentary Assessment (EMA) Smoking Log

Over the 7-day study period, whenever participants were inclined to smoke, they were requested to answer a series of self-administered open ended questions in real time, including their emotion, activity, and the time of day while smoking. Additionally with each cigarette, they rated their urge to smoke (1 lowest –10 highest), whether they were alone (yes/no), and if they had any alcohol in combination with the smoking session. Participants were told to smoke ad lib.

Self-reported emotion and activity data were coded independently by two research associates and assessed for inter-rater reliability with an agreement rate of 96.3% and 99.3%, respectively. Stated activities were categorized as either leisure, performative, social, or interactive.²¹ Emotions were categorized as either positive, negative, neutral, or mixed when respondents reported a mixture of both negative and positive emotions. Emotions were categorized as follows: positive responses include excited, content, happy, and having fun; examples of negative responses include anxious, annoyed, fearful, disappointed and depressed; examples of neutral responses include fatigued, pensive, resting, tired; examples of mixed responses include tired/happy, contemplative/worried, bored/cheerful and calm/tired/bored. Activities were categorized as follows: leisure activities, after dinner, watching TV, at home, on the computer, and drinking coffee; examples of performative activities, defined as an activity aimed at reaching goals or fulfilling tasks, include, completing chores, driving, getting child ready for school, getting ready for work/school, and multitasking; examples of social occasions, defined as an activity that combines leisure and interaction, include at a party, at a friend’s home, at the casino, and drinking at a bar or restaurant; examples of interaction, defined as an activity with dialogue, include talking on the phone, arguing with another, business calls, and having a conversation.

2.3.3. Daily log

On each of the 7 days participants filled out a daily log (regardless if they smoked or not) with a set of questions on how much of the time they felt the urge to smoke, how strong the urges to smoke have been²², and their overall stress level for the day (1 lowest –10 highest).

2.4. Statistical Analysis

Frequencies and proportions were computed for categorical variables and means with standard deviations were calculated for continuous variables. We further explored differences between converted ITS and native ITS by demographics, smoking history and smoking dependency using Independent two-sample t-tests, chi-squared test, Fisher’s exact test or Mantel-Haenszel trend test as appropriate for the type of variables. Independent two-sample t-tests were used to compare the subscales and overall temptation of the Situational Temptation Scale for Smoking between the two groups. Fisher’s exact tests were used to measure differences in activity, emotion, and time of day while smoking. Mixed effects logistic regression was used to model the repeated measurements of binary outcomes such as being alone during smoking sessions on the two groups (native ITS vs. converted ITS). Ordinal logistic regression with mixed effects was used to model the repeated measurements of ordinal outcomes such as the urge to smoke during smoking sessions. All statistical analyses were done using SAS software version 9.4 (SAS Institute, Cary, NC, USA) or R programming language version 3.3.2 (R Foundation). All statistical tests were two-sided and the significance level (alpha) used was set at 0.05.

3. RESULTS

3.1 Baseline questionnaire assessments

The study included 60 smokers, including 36 converted ITS and 24 native ITS. There were no significant differences between converted ITS and native ITS by gender, age, race, education level, and income (Table 1). Combined, the groups reported smoking intermittently for an average of 8.14 years (Standard deviation (SD) = 7.38). When comparing years as intermittent smokers, native ITS smoked almost double the number of years as compared to converted ITS, averaging 11.42 (SD=6.71) and 5.95 (SD=7.06) years respectively (p <0.01). Native ITS trended towards smoking fewer days in the past 30 days than converted ITS (12.33 vs. 14.90, P=0.06, respectively) however, both groups reported an average of 3 cigarettes per day (CPD) on days smoked. The converted ITS smoked 11.44 (SD=7.77) CPD while they were a daily smoker, before switching to smoking intermittingly 6.50 (SD=6.96) years ago. Native ITS initiated smoking later in life compared to converted ITS smokers (21.38 vs.18.22; P=0.02, respectively). Native ITS were less likely to ever try to make a quit attempt compared to the converted ITS (P<0.01). Furthermore, only 13% of all the ITS ever received counseling for smoking, and none of the native ITS reported ever receiving counseling.

Table 1.

Demographics and smoking behavior among native and converted intermittent smokers

	Total (N=60)	Native Intermittent Smokers (N=24)	Converted Intermittent Smokers (N=36)
	N (%) / Mean (SD)	N (%) / Mean (SD)	N (%) / Mean (SD)	p-value
Demographics
Gender (male)	29 (48.33%)	13 (54.17%)	16 (44.44%)	0.46b
Race				0.62d
White	47(78.33%)	20 (83.33%)	27 (75%)	─
Black	8(13.33%)	3 (12.5%)	5 (13.89%)	─
Asian	3(5%)	0 (0%)	3 (8.33%)	─
Other	2(3.33%)	1 (4.17%)	1 (2.78%)	─
Age	34.86 (8.90)	34.22 (8.77)	35.29 (9.09)	0.66a
Education				0.38c
High School Graduate or Less	11(18.33%)	5 (20.83%)	6 (16.67%)	─
Some College/Associate Degree	27(45%)	12 (50%)	15 (41.67%)	─
Bachelor’s Degree or Higher	22(36.67%)	7 (29.17%)	15 (41.67%)	─
Annual Income ($1,000s)	49.41(38.56)	40.77 (29.77)	54.94 (42.73)	0.17a
Smoking History
Total Years as ITS	8.14 (7.38)	11.42 (6.71)	5.95 (7.06)	<0.01a
Previous years as a Daily Smoker	─	─	8.90 (7.31)	─
Years Since Switching	─	─	6.50 (6.96)	─
Number of smoking days in past 30 Days	13.88(5.25)	12.33 (5.43)	14.90 (4.93)	0.06a
Self-Reported CPD on days smoked	3.23(1.83)	3.40 (2.32)	3.13 (1.43)	0.61a
CPD when smoked daily	─	─	11.44 (7.77)	─
Lived with other smokers before age 18 (Yes)	43(74.1%)	16 (69.6%)	27 (77.1%)	0.52b
Smoking to cope with stress				0.051c
Not really	9(15.0%)	7 (29.2%)	2 (5.6%)	─
A little	15(25.0%)	5 (20.8%)	10 (27.8%)	─
Somewhat	22(36.7%)	8 (33.3%)	14 (38.9%)	─
Very much	14(23.3%)	4 (16.7%)	10 (27.8%)	─
Ever felt addicted to tobacco (Yes)	33(55%)	8 (33.3%)	25 (69.4%)	<0.01b
Ever felt like you needed a cigarette (Yes)	54(90%)	20 (83.3%)	34 (94.4%)	0.21d
Age of smoking initiation	19.48 (5.27)	21.38 (4.83)	18.22 (5.24)	0.02a
Ever tried to quit smoking (Yes)	37 (62.7%)	8 (34.8%)	29 (80.6%)	<0.01d
Ever received counseling for smoking (Yes)	8 (13.3%)	0 (0%)	8 (21.6%)	0.02d
Dependence
FTND				0.20c
Very low dependence (0–2)	54 (90.0%)	23 (95.8%)	31 (86.1%)	─
Low dependence (3–4)	5 (8.3%)	1 (4.3%)	4 (11.1%)	─
Medium dependence (5)	1 (1.7%)	0 (0%)	1 (2.8%)	─
Mean FTND score	0.68 (1.17)	0.46 (0.78)	0.83 (1.36)	0.18a
PSCDI				0.44c
No dependence (0–3)	41 (68.3%)	18 (75.0%)	23 (63.9%)	─
Low dependence (4–8)	18 (30%)	6 (25.0%)	12 (33.3%)	─
Medium dependence (9–12)	1 (1.7%)	0 (0%)	1 (2.8%)
Mean PSCDI score	2.9 (2.03)	2.21 (1.69)	3.36 (2.13)	0.30a
Mean HONC score	4.07 (2.09)	2.91 (1.59)	4.81 (2.05)	<0.001a
Time to first cigarette				0.05c
Within 5 minutes	2 (3.3%)	0 (0%)	2 (5.6%)	─
6−30 minutes	4 (6.7%)	0 (0%)	4 (11.1%)	─
31−60 minutes	2 (3.3%)	1 (4.2%)	1 (2.8%)	─
After 60 minutes	52 (86.7%)	23 (95.8%)	29 (80.6%)	─
Mean time to first cigarette (min)	424.3 (286.69)	460.0 (271.9)	400.6 (297.5)	0.44a

Abbreviations: ITS; Intermittent smoker, CPD; cigarettes per day, FTND; Fagerstrom Test for Nicotine Dependence, HONC; Hooked on Nicotine Checklist, PSCDI; Penn State Cigarette Dependence Index.

Statistical test:

^atwo-sample t-test;

^bChi-Square test;

^cMantel-Haenszel trend test;

^dFisher’s exact test

For both groups combined, 74% experienced living with a smoker prior to the age of 18. The majority of both groups reported the feeling of really needing a cigarette, yet only 33% of native ITS ever felt like they were addicted to tobacco compared to 69% of the converted ITS group (P<0.01). Additionally, converted ITS reported that smoking helped them cope with stress more often than native ITS, with 67% compared to 50% responding somewhat or very much (p <0.005).

As shown in Table 1, both groups had very low levels of nicotine dependence as determined by the FTND (native ITS: 0.46 vs. converted ITS: 0.83; P=0.18) and PSCDI (native ITS: 2.21 vs. converted ITS: 3.36; P=0.30). Converted ITS score significantly higher on the HONC, a measure of loss of autonomy (native ITS: 2.91 vs. converted ITS: 4.81; P<0.001). Native ITS waited 60 minutes longer than converted ITS to smoke their first cigarette of the day, although the difference was not significant (native ITS: 460 mins vs. converted ITS: 400 mins; P=0.44).

Results for the Situational Temptation for Smoking Inventory and its individual domains are shown in Table 2. Overall, participants indicated that smoking for reasons of habituation/addiction were the least important situational factors for smoking. The subscales that assessed situations with negative affect or positive/social scenarios prompted the most temptations to smoke. Among the two groups, there were lower scores for native ITS for the negative affect and social/positive subscales and the overall temptation score compared to converted ITS, but the differences were not significant. Scores were significantly lower for the habitual/addictive subscale (e.g. ‘when needing a lift’ or ‘first thing in the morning’) in the native ITS group (P=.02).

Table 2.

Situation Temptation for Smoking Inventory

	Total (N=60)	Native Intermittent Smokers (N=24)	Converted Intermittent Smokers (N=36)
	Mean (SD)	Mean (SD)	Mean (SD)	p-value
Negative Affect	3.1 (1.06)	2.81 (1.34)	3.13 (.83)	.31
Social/Positive	2.88 (.67)	2.85 (.64)	2.9 (.69)	.79
Habitual/Addictive	1.71 (.64)	1.48 (.47)	1.85 (.71)	.02
Overall Temptation	2.62 (.53)	2.48 (.54)	2.7 (.51)	.12

Statistical test: two-sample t-test. Scoring: 1=Not At All Tempted to 5 = Extremely Tempted

3.2 EMA Smoking Log

A total of 574 smoking sessions were recorded over a 7 day period. The native ITS had a total of 198 (34%) smoking sessions and the converted ITS had a total of 376 (66%). The groups reported smoking more often when under positive emotional stimuli (47%) than negative (32%), neutral (16%), or mixed (5%), and during activities of leisure (48%) than during performative duties (29%), social (16%) or interactive occasions (7%). Both groups combined smoked more during the evening (59%) than during the afternoon (24%) or morning (17%). There were no significant differences in emotions, activities, or time of day while smoking between the native ITS and converted ITS (Figure 1). In a mixed effects logistic regression analysis, converted ITS were more likely to smoke alone (71%) compared to the native ITS (36%) [odds ratio, 6.82; 95% confidence interval, 2.17–21.66; p <.001].

Figure 1.

Comparison of ecological momentary assessment smoking log among native (n=24) and converted (n=36) intermittent smokers

Among the urges prior to smoking, 63% of total smoking session were prompted by an urge of 5 or less and 37% were prompted by an urge of 6 or more (Figure 2). Converted ITS showed a higher urge level to smoke overall compared to native ITS (odds ratio, 1.92; 95% confidence interval, 0.93–3.99; P=.072). When comparing the urge of the first cigarette of the day to the rest of the cigarettes smoked in the day, there was no significant difference in urge (converted ITS: odds ratio, 0.97; P=.88; native ITS: odds ratio, 0.86; P=.48).

Figure 2.

Distribution of urge to smoke during smoking sessions among native (n=24) and converted (n=36) intermittent smokers

3.3. Daily log

Overall for both groups combined, there was a higher mean stress level on days smoked (4.37, SD=2.54) than for non-smoking days (3.06, SD=2.06, P<0.0001). Stress levels were similar between the two ITS groups on smoking days (P=0.86) and non-smoking days (P=0.87).

On smoking days, ITS reported having smoking urges either some of the time or greater during 56% of the time and reported the strength of the urges to be moderate or greater 60% of the time. There were no significant differences between the two groups on strength of urges (P=0.45) or the amount of time feeling the urge to smoke (P=0.11) on smoking days.

On non-smoking days, ITS reported having smoking urges some of the time or greater only 12% of the time and reported those urges to be moderate or greater 11% of the time. There was no significant difference between the two groups on strength of urges (P=.65). There was a significant difference in how much of the time they felt the urge to smoke during the day with the converted ITS feeling the urge to smoke more than native ITS (odds ratio, 1.68; 95% confidence interval, 1.62–1.75; P<0.0001).

4. DISCUSSION

The current study examined environmental and behavioral factors observed in real-time smoking sessions among converted and native ITS. The socio-demographic characteristics and levels of tobacco dependence were similar between the two groups. For example, both groups reported a similar level of CPD on days smoked, but converted ITS smoked a greater number of days in the past month. Nearly all subjects in our study had very low levels of nicotine dependence as determined by mean scores of less than one on the FTND and on similar scales. Despite being very low, dependence scores were slightly higher in converted versus native ITS. Only 17% of converted ITS smoked their first cigarette within 30 minutes after waking, whereas 0% of native ITS reported doing so. ITS were previously reported to smoke more often in the evening and night.²³ Almost 70% of converted ITS and only 33% of native ITS reported ever feeling addicted to tobacco, yet the level of addiction/habituation reported in the temptation scale was quite low. It should be noted that the ITS in the current study differ from “social smokers”, a term used mostly to characterize ITS in college settings and young adults, who typically smoke when drinking, with friends, and to relieve boredom.²⁴

Similar to daily smokers, about 75% of the ITS in our study lived with a smoker when growing up. Parental smoking is a predictor of nicotine addiction onset and tobacco use. ²⁵ Twin studies suggest heritable factors as well as environmental factors influence adult smoking behaviors. ²⁶ Despite the low levels of dependence and relative infrequent smoking in ITS, they share risk factors for smoking with daily smokers.

Tobacco dependence is a multi-faceted behavior. While dependence measures such as time to first cigarette and FTND are excellent measures of nicotine exposure, they may not necessarily be strongly correlated with the desire to smoke. Theoretical and empirical work indicates that the need and urge to smoke are indicators of the severity of dependence and quit success.^27,28 For those with low dependence, the urge to smoke is more related to psychological addiction. Most ITS smoking sessions in the current study were characterized by moderate urges to smoke on a 10-point scale.

Data collected from the Situational Temptation Scale for Smoking showed that converted ITS compared to native ITS had similar temptations to smoke when confronted with situations related to negative affect (i.e., when in conflict, frustrated, angry) and social/positive affect (i.e., when happy, with friends/family), but the converted group showed higher temptations related to habit/craving compared to native ITS. Others have found converted ITS and native ITS have similar cue reactivity.²⁹ It’s possible that converted ITS are more sensitive to cravings and habitual smoking because they were previously daily smokers, making their connection with cigarettes harder to break. In our study, converted ITS were more likely to have ever made a past quit attempt and sought counselling for smoking cessation compared to native ITS. This is consistent with other findings¹⁰ and further provides evidence that some converted ITS may be transitioning from daily smoking to quitting. In comparison to daily smokers, ITS have overall lower temptations to smoke according to a survey of 2,921 adult daily smokers, using a modified version of the Situational Temptation Scale for Smoking.³⁰ The temptation to smoke may be an important predictor in understanding the effects of non-nicotinic factors in cigarette dependence among subgroups of smokers.

Perceived temptation surveys are easily distributed that can quickly assess smoking motives; however, EMA data provide a more realistic measure of smoking behavior. Data collected from the EMA smoking log showed that converted and native ITS overall smoked most frequently during leisure activities and while in a positive emotional state, most often with low to moderate urges to smoke. Both groups consumed cigarettes more frequently in the evening than during the afternoon or morning. There were some suggestive differences, in how their emotion and environment motivated them to smoke. In addition, 37% of native ITS total smoking sessions happened while they were alone and most of the smoking sessions happened without alcohol (67%) [Figure 1].

Given that the majority of smoking sessions were initiated by a moderate urge level and the smokers in this sample experience minimal levels of dependence to none at all, it further suggests that non-nicotinic factors act as the strongest incentive for this group to smoke compared to nicotine seeking factors. A previous study theorized ITS to be “peak seekers,” in that they seek the pharmacological effects of nicotine under certain contexts, but do not need regular nicotine intake as daily smokers do.^23,31 EMA data show that ITS are much more likely than daily smokers to be prompted by environmental and/or psychological cues or triggers to smoke, which may be positive or negative, and continue to smoke because it serves as a compliment when paired with other stimuli, such as drinking, with friends, while driving, or relaxing²³. However, certain environmental factors can be ruled out as smoking triggers for ITS such as morning coffee, because of the long time between waking and the first cigarette.³² In trial data, smoking cues have been associated with negative affect, which is consistent with our EMA findings.³² Although the current study did not capture self-reported reasons for smoking (e.g. feeling more confidence, facilitating social exchange etc.), ITS report fewer reasons for smoking than daily smokers.³³ Recently, Shiffman and Terhorst (2017)³⁴ showed ITS smokers experience sensory rewards from smoking like daily smokers do^35,36, but reported more aversion than daily smokers providing a possible explanation for continued smoking without daily consumption. Our data did not specifically capture sensory experiences during smoking sessions, but future work should be done to further these findings between native and converted ITS.

The native ITS were less likely to report making a quit attempt and never received counseling for smoking, which is consistent with data from the 2003 Tobacco Use Supplement to the Current Population Survey.¹⁰ This is concerning given the fact the native ITS in our sample have been smoking for an average of 11 years. Additionally, the native ITS started smoking later in life (~21 years old). Implications for these findings include better identification of ITS within medical practice to start the process of smoking cessation even among nondaily smokers and targeting young adults beyond the teen years about the dangers of smoking initiation. One positive finding was the converted ITS group not only switched from being a daily smoker; they also reduced from 11 to 3 CPD. This is a considerable decrease in smoke exposure and consequently reduces harm, albeit there is still no safe level of smoking.

Our analysis contains some limitations which should be noted. First, we did not assess whether participants smoked intermittently because of aversive symptoms associated with smoking or because of current medications or illnesses that precluded them from smoking more regularly. Although our analysis consisted of over 500 smoking sessions, the relatively small participant sample size should be considered when interpreting findings. Nonprobability sampling methods were used to capture our sample; therefore, the degree of generalizability may be diminished. Additionally, our sample was predominantly white, whereas a greater proportion of ethnic minorities than whites are ITS.³⁷ We used a single item to assess the urge to smoke, although this has shown to have the same predictive value of several commonly used scales of smoking urges.³⁸ Our baseline survey could be subject to recall bias. Lastly, while allowing participants to answer open-ended prompts on the smoking log generated authentic responses in real time, it required post-hoc grouping for the analysis. However, the inter-rater reliability of grouping open-ended responses showed high concordance between the two raters.

Consistent with most work, ITS have very low levels of dependence. Their intermittent smoking patterns over long periods of time still put this population at an increased health risk. Understanding why they continue to smoke and what reinforcing factors are the strongest among the ITS subgroups may be important for developing effective smoking interventions for this group.

Scientific Significance.

The current paper identifies environmental and behavioral factors that are associated with smoking among ITS in real time.