The Impact of Coping With Stressful Events on Negative Affect and Cravings Among Smokers With Mood Disorders

Abstract

Introduction

Smokers with mental illness report elevated levels of stress and negative affect. Craving is often cited as a key precipitant of smoking. Coping with stress has been associated with reduced cravings among smokers attempting to quit. However, the effect of coping with stress on negative affect and craving among smokers with mental illness is not well understood. This study investigated whether coping with stress predicts lower subsequent craving, mediated by reduced negative affect, among socioeconomically disadvantaged smokers with mood disorders.

Aims and Methods

This study used ecologically momentary assessment (EMA) data from a randomized controlled trial involving smokers with mood disorders. The final sample included 39 participants.

Results

Traditional mediation path analyses showed that coping with stress predicts lower craving (p = .02) through its impact on negative affect (p < .001) for the contemporaneous model (ie, when craving was measured at the same report as coping). However, coping with stress did not have a prospective effect on craving (ie, when craving was measured at the next report, up to 12 hours later) (p = .11).

Conclusions

The results suggest that coping with stress reduces craving through negative effect, but only for a limited timeframe. The findings could guide future research on the length of time that the effect of coping lasts and research on interventions to increase coping with stress among smokers with mental illness.

Implications

This is the first study to use EMA to demonstrate that coping with stressful events effectively reduces craving through reducing negative affect among smokers with mood disorders. This finding suggests that individuals heavily burdened with stress and negative affect benefit from coping with stress. We utilized within-subject analyses of EMA data which allowed us to understand these effects within an individual near real time. Our sample is hard to reach and ethnoculturally diverse. Findings could guide intervention research on helping smokers with mental illness cope when experiencing stress.

Introduction

Cigarette smoking is the leading cause of preventable deaths in the United States.¹ Individuals with mental illness are over three times more likely to smoke, and smokers with mental illness have greater tobacco-related mortality compared with smokers without mental illness.^1,2 Even though smokers with mental illness are as motivated to quit as smokers in the general population,³ they are 25% less likely to quit smoking.² Understanding the dynamic relationships between factors that contribute to difficulties in quitting smoking among persons with mental illness (such as stress, negative affect, and craving^4,5) is important, as smokers with mental illness are burdened with high levels of stress and negative affect.^4–6

Approximately 90% of smokers with mental illness report stress as a reason for smoking⁶ and many smokers report stress as one of the biggest challenges in quitting smoking.⁷ Ad lib smokers and smokers attempting to quit both report that they smoke or lapse to reduce their stress levels⁵ and negative affect.⁵ Compared with individuals without mental illness, individuals with mental illness report greater stress levels and negative affect,⁸ fewer coping resources,⁹ and more severe withdrawal symptoms during a quit attempt.¹⁰ Since these constructs fluctuate rapidly, studies have utilized real-time data such as ecological momentary assessment (EMA) to capture these dynamic changes.¹¹ For example, EMA studies showed that negative affect rapidly increased hours before the lapse and spiked during the 6-hour period before the lapse,⁸ and experiencing stress predicted increases in negative affect in the context of quitting smoking.¹² Additionally, an EMA study among smokers with post-traumatic stress disorder (PTSD) where around half were also diagnosed with a depressive disorder found that PTSD symptom severity assessed at each EMA report predicted negative affect at the subsequent report.¹³ This suggests that those with psychiatric disorders may be more vulnerable to experiencing greater negative affect during a quit attempt. While stress and negative affect have been linked with smoking, there are limited EMA studies on how coping with stress affects smoking particularly among individuals with mental illness.

Active coping with stress is the effort to manage demands that are seen as taxing through problem solving strategies, unlike avoidant coping which utilizes withdrawal or escape from the problem.¹⁴ In population surveys, smokers with mental illness reported that lack of active coping with stress is a contributing factor to ad lib smoking.¹⁵ Lack of coping during stressful situations has also been shown to be a predictor of relapse after smoking cessation.^16,17 EMA studies among people without mental illness have yielded mixed findings on the effect of coping on smoking outcomes. One study found that coping with stress prevented lapses through reducing cravings,¹⁸ while another study did not find an effect of coping on smoking risk.¹⁷ However, studies have revealed that coping with stress predicts reduced negative affect and increased positive affect,¹⁹ and has also been associated with lower cravings.²⁰ Furthermore, one EMA study found that dispositional mindfulness (which could be characterized as active coping) predicted lower negative affect among smokers attempting to quit.²¹ These findings highlight the benefits of coping with stress on negative affect and cravings, but these effects have not been examined among smokers with mental illness.

Craving for a cigarette is characterized as a desire or urge to smoke²² and can be considered a core feature of addictive behavior.²³ In fact, 90% of daily smokers reported having at least some craving when they have not smoked for a few hours.²⁴ One study found that 36% of smokers with a comorbid mental health disorder and a substance use disorder reported craving as the primary reason for smoking.²⁵ EMA studies among smokers attempting to quit have found that higher duration and frequency of cravings predict smoking lapses and relapses,^24,26 with ten out of 66 EMA studies in a review demonstrating significant associations between craving and smoking in the following minutes.²⁷ Another EMA study found that craving predicted substance use (including tobacco use) among individuals with and without psychiatric disorders, but craving intensity was higher among individuals with mood disorders.²⁸ This suggests that cravings might be especially challenging for individuals with psychiatric disorders.

Negative affect also predicts increased craving.^5,29 A study among participants in inpatient detox treatment found that those with elevated negative affect had higher cravings for cigarettes.³⁰ A randomized controlled laboratory study among daily smokers not receiving treatment revealed that participants who received a negative affect script experienced greater craving, compared with participants who received a neutral script.³¹ EMA studies have also shown that individuals experienced greater craving when they were experiencing higher negative affect (which fluctuated within individuals).³² Additionally, an EMA study found that daily hassle predicted cravings and this relationship was stronger among smokers with low distress tolerance.³³ However, no EMA studies have investigated negative affect and craving in the context of smoking cessation among persons with mental illness, a population challenged with low distress tolerance. Additionally, there is scant research on the impact of coping with stress on craving.

In the current study, we utilized EMA data from a smoking cessation trial in the context of a quit attempt (ie, over the course of 28 days post-target quit date) to investigate the impact of active coping with stress on negative affect and craving. We hypothesized that active coping with a stressful event will reduce the intensity of subsequent craving. We also predicted that the relationship between coping and craving will be mediated by a reduction in negative affect. We used the traditional mediation path³⁴ to investigate our hypotheses.

Methods

Participants

This study used data from a randomized controlled smoking cessation study for adult daily smokers with mood disorders. Participants were recruited at an outpatient psychiatric facility in the Bronx, NY. Participants were adult daily smokers receiving treatment for a depressive or bipolar disorder who intended to quit smoking. The inclusion criteria included smoking at least five cigarettes per day over the past 6 months, intent to quit smoking, and receiving treatment for a depressive or bipolar disorder at the psychiatric clinic. Participants were excluded if they were pregnant or breastfeeding, used other tobacco products or marijuana more than four times a week, used pharmacotherapy for smoking cessation, or they had acute psychiatric symptomology, cognitive impairment, cardiovascular disease, a psychotic disorder, or a nonnicotine substance use disorder within the past 6 months.³⁵ Participants were randomized to either receive a smartphone-assisted mindfulness smoking cessation intervention with contingency management (SMI-CM), or enhanced standard treatment plus noncontingent incentives (EST).³⁵ Participants in the both conditions received counseling sessions, but the counseling sessions in the EST condition lacked a mindfulness component. Participants in the SCI-CM condition received incentives for submitting expired CO monitor videos, whereas participants in the EST condition received incentives at a random schedule noncontingent on the CO levels.

Final Sample

For the current study, we included participants who reported experiencing at least one stressful event after their quit date (described below). Our final sample consists of 39 (80% out of 49 enrolled) participants. There were no significant demographic or baseline differences between the included and excluded individuals (ie, individuals who did not experience stressful events) (Table 1).

Table 1.

Demographic Characteristics of Included Sample (N = 39) and Excluded (N = 10)

Characteristics	Retained cases (n = 39)	Excluded cases (n = 10)	t or χ 2	p
Female	30 (76.9%)	7 (70.0%)	.21	.65
Latinx	25 (64.1%)	6 (60.0%)	.06	.81
Race/ethnicity			3.07	.55
White	18 (46.2%)	3 (30.0%)
Black/African American	10 (25.7%)	5 (50.0%)
Native Hawaiian/Pacific Islander	1 (2.6%)	0 (0.0%)
American Indian/Alaskan Native	2 (5.1%)	1 (10.0%)
Highest level of education			1.90	.75
Four-year college	2 (5.1%)	0 (0.0%)
Some college or technical school	14 (35.9%)	4 (40.0%)
High school graduate/GED	8 (20.5%)	3 (30.0%)
Some high school	11 (28.2%)	3 (30.0%)
Less than high school	4 (10.3%)	0 (0.0%)
Employment			6.10	.41
Unable to work/disabled	18 (46.2%)	5 (50.0%)
Out of work for more than 1 year	7 (17.9%)	2 (20.0%)
Employed for wages (full-time)	2 (5.1%)	0 (0.0%)
Employed for wages (part-time)	2(5.1%)	2 (20.0%)
Retired	3 (7.7%)	0 (0.0%)
Student	1 (2.6%)	1 (10.0%)
Household income			.14	.71
$0–24 999	29 (74.4%)	8 (80.0%)
$25 000–49 999	10 (25.6%)	2 (20.0%)
	M (SD)
Age	49.4 (11.0)	46.6 (20.2)	.45	.50
Current/history of bipolar diagnosis	20 (51.2%)	6 (60.0%)	.24	.62
Current/history of depressive disorder diagnosis	19 (48.7%)	4 (40.0%)	.24	.62
Currently on Wellbutrin (not for smoking)	9 (23.1%)	1 (10.0%)	.84	.36
Nicotine dependence (FTCD)	5.0 (1.7)	4.4 (2.1)	−1.03	.31
Years smoking	30.2 (12.0)	26.5 (14.9)	.66	.42
Cigarettes per day (past 7 days)	10.6 (4.9)	9.7 (4.5)	.30	.59
Number of past serious quit attempts (>24 h)	2.7 (3.1)	2.1 (1.3)	.39	.54
Depressive symptoms (PROMIS-D8a)	22.1 (8.7)	23.9(10.2)	.57	.57
Anxiety symptoms (PROMIS-A8a)	24.1 (8.7)	26.8 (8.7)	.87	.38

FTCD = Fagerström Test for Cigarette Dependence.

Procedures

In the parent study, the primary aim was to test the effectiveness of a mindfulness-based intervention consisting of two individual counseling and two brief phone sessions as well as mindfulness practice through a smartphone.³⁵ In the control condition, participants also received two individual counseling and two brief phone sessions, but without the mindfulness practice. All participants completed daily EMA reports.

Measures

Participants completed self-report, valid, reliable measures at their baseline assessment.³⁰ The Fagerström Test for Cigarette Dependence (FTCD) was used to measure nicotine dependence.³⁶ Psychiatric symptoms were measured by PROMIS—depression and anxiety short forms 8a.³⁷

Ecological Momentary Assessment

Participants in both conditions were prompted to complete EMA reports pseudo-randomly (at least 1.5 hours apart during waking hours) five times per day for 38 days. Each report took between three to five minutes to complete. Participants received a compensation of $0.25 per report completed, with a bonus of $10 per week for completing at least 90% of the reports, with a maximum compensation of $88. The following measures were assessed by EMA.

Experience of and Coping With Stressful Events

The EMA reports assessed if participants experienced a stressful event(s) (Yes/No) since their previous EMA report. If participants endorsed “Yes,” then they were asked “How did you cope with it?” with the following options: (1) Accepting/Mindfulness, (2) Smoking, (3) Alcohol/Substance, (4) Talk to someone, (5) Thoughts (Distracting/Helpful), (6) Behaviors (Distracting/Helpful), (7) Other, or (8) Did not try to cope. The options that reflect active coping with stress,³⁸ “Accepting/Mindfulness,” “Talk to someone,” “Thoughts,” and “Behaviors,” were coded as coping. The responses “Did not try to cope” as well as “Smoking” and “Alcohol/Substance” were coded as not coping with stress, as substance use is considered an avoidant coping style³⁸ and not active coping. Participants could select more than one coping method. If participants reported utilizing both active coping and substance use, we coded that response as not coping with stress because substance use is an avoidant coping style. “Other” (5.9% of responses) was fit into one of the categories depending on the written response.

Negative Affect

Negative affect was measured through four items (“sad or depressed,” “nervous or anxious,” “angry,” “ashamed or guilty”) based on similar items found in Wisconsin Smoking Withdrawal Scale (WSWS)³⁹ (“sadness,” “anger,” “anxious”) and Positive and Negative Affect Schedule (PANAS)⁴⁰ (“ashamed,” “guilty,” “nervous”) in order to capture a broad range of negative affect. These items were rated on a five-point scale ranging from 1 (not at all) to 5 (extremely). The correlations among these items ranged from .34 to .51 (ps < .01). Similar EMA measures of negative affect have been validated by EMA studies demonstrating that negative affect is spiked after a stressful event⁴¹ or at the outset of a quit attempt.⁴²

Craving

Craving was measured by asking “How are you feeling RIGHT NOW? I’ve been bothered by the DESIRE/URGE to smoke a cigarette” with a Likert scale of 1–5, 1 being “Not at all” and 5 being “Extremely.” Research indicated that craving assessed in the moment is less prone to recall bias given its rapid fluctuations.⁴³

In the EMA report, the momentary craving and affect items were asked first, followed by the stressful event and coping questions on a subsequent page which reduced the likelihood that responses to the momentary craving and affect items were affected by the stressful events/coping questions.

Participants also recorded the total number of cigarettes they smoked since their last EMA report. For the binary variable of smoking status, one or more cigarettes was coded as smoking.

Analysis Plan

We conducted analyses using the traditional mediation path model (ie, causal steps approach),³⁴ the most widely used method in assessing mediation in psychology studies.⁴⁴ We did not utilize formal mediation analyses using indirect effects model as we did not have a large enough sample size and this model is not commonly used for within-subject analyses.⁴⁵ We used a multilevel model with EMA reports (Level 1), nested within individuals (Level 2). Three sets of models were fit to test our hypothesis (Figure 1). Smoking status was included as a covariate in all models, as research shows that craving ratings are more likely to be the lowest immediately after having smoked.⁴⁶ For all paths, only the intercept was allowed to vary across individuals as specifying other parameters in the model as random did not improve model fit.

Figure 1.

Conceptual model of hypothesized relations between coping, negative affect, and craving (Hypotheses 1a, 1b, and 1c). Three separate models were run to test the hypothesized effects of coping (between t₋₁ and t₀) on craving (t₀) (c path), coping on negative affect (a path), and negative affect (t₀) on craving (t₀) (b path), controlling for smoking status (between t₋₁ and t₀). Models were rerun for prospective analyses on the effects of coping (between t₋₁ and t₀) on craving (t₁) (c path), coping on negative affect (a path), and negative affect (t₀) on craving (t₁) (b path), controlling for smoking status (between t₀ and t₁).

Contemporaneous Model

For the first model, the direct effect of the causal variable (coping) between the last EMA report (t₋₁) and the index report (t₀) on the outcome variable (craving) at t₀, was tested, controlling for smoking status (between t₋₁ and t₀) (c path in Figure 1). For the second model, we tested the effect of the causal variable (coping) between t₋₁ and t₀ on the mediator (negative affect) at t₀, controlling for smoking status (between t₋₁ and t₀) (a path in Figure 1). For the third model, we tested the effect of the mediator (negative affect) at t₀ on the outcome variable (craving) at t₀, controlling for smoking status and coping (between t₋₁ and t₀) (b path in Figure 1). We controlled for coping in the b path, as the mediator and outcome might both be affected by the causal variable.⁴⁵

Prospective Model

The same models were prospectively tested using the levels of craving reported at the next EMA report (t₁) completed within 12 hours following the index report (Figure 1). This was to investigate whether the effect of coping on craving lasts for an extended period of up to 12 hours. For the c path, the effect of coping (between t₋₁ and t₀) on craving at t₁ was tested, controlling for smoking status (between index to t₁). The a path was identical to the concurrent model. For the b path, the effect of negative affect (t₀) on craving at t₁ of the index report was tested, controlling for smoking status (between index report to t₁). Regression coefficients for predictors were allowed to vary across individuals as long as doing so significantly reduced residual sum of squares, suggesting a better model fit.⁴⁷

We conducted within-subject centering for all the predictor variables.⁴⁸ It is important to distinguish the within-subject effects (EMA report level) from the between-subject effects (individual level) as some individuals may consistently have distinct behavior from others. The individual’s mean value is subtracted from each observation value, creating the within-subject value. The between-subject value is the individual’s mean value across all observations. For all predictor variables, we included both the within-subject variable and the between-subject variable.

During the 28-day postquit assessment period, participants provided 3344 EMA reports (average of 85.7 reports per person, or 3.1 out of five reports per day). We conducted the analyses using all EMA reports with stress (420 reports). The proportion of EMA reports with stress (12.6%) was higher than seen in a previous study among smokers without mood disorders where stressful events were reported in 8% of EMA reports during a quit attempt,¹⁹ suggesting that smokers with mood disorders experience stressful events more frequently than the general population in the context of quitting smoking. The index reports had an average time of (M = 297.6 minutes, SD = 269.9) from the previous report, and an average of (M = 293.3 minutes, SD = 257.6) to the next report. We also conducted analyses with: (1) reports with a previous report within 12 hours (351 reports) and (2) reports with a previous report on the same day (342 reports). However, the findings remain unchanged, thus we report the analyses using all EMA reports with stress to include a larger number of reports. We also conducted analyses including demographic variables (ie, age, race/ethnicity, income, education, employment, nicotine dependence) and intervention condition as covariates. The findings remained unchanged, thus we present the models without the demographic variables and intervention condition for parsimony.

Results

Sample Characteristics

As detailed in Table 1, 76.9% of the study sample was female, and 74.4% were socioeconomically disadvantaged. Participants were ethnoculturally diverse. The mean age was 49.4 and around half were diagnosed with bipolar disorder.

Coping and Craving (c Path)

Higher within-subject coping predicted lower craving (p = .02), while there was no significant effect on craving for between-subject coping or smoking status (Table 2). This suggests that whether an individual coped with a specific stressful event, not the individual differences in coping tendency (ie, general likelihood/frequency of coping), affected their subsequent craving.

Table 2.

Trimmed Hierarchical Linear Model (HLM) Analysis of the Effects of Coping (Between t₋₁ and t₀) on Craving (t₀), Coping (Between t₋₁ and t₀) on Negative Affect (t₀), Negative Affect (t₀) on Craving (t₀)

Predictor	Coefficient	95% CI	Approx. df	p
Direct effect c path (index report)
Smoking (since t−1 to t0: Y/N)	.01	[−.24, .25]	377	.95
Coping (between t−1 and t0)
Within-subject coping	−.28	[−.52, −.05]	377	.02*
Between-subject coping	−.68	[−1.7, .33]	37	.18
a path (index report)
Smoking (since index to t1: Y/N)	−.07	[−.23, .09]	377	.37
Coping (between t−1 and t0)
Within-subject coping	−.29	[−.45, −.14]	377	<.001*
Between-subject coping	.32	[−.34, .99]	37	.33
b path (index report)
Smoking (since t−1 to t0: Y/N)	.03	[−.19, .28]	376	.71
Negative affect (index)
Within-subject negative affect	.47	[.32, .62]	376	<.001*
Between-subject negative affect	.67	[.21, 1.13]	36	.01*
Coping (between t−1 and t0)
Within-subject coping	−.14	[−.37, .09]	376	.22
Between-subject coping	−.90	[−1.84, .03]	36	.06

CI = confidence interval.

*p < .05

There was no significant effect of either within-subject coping or between-subject coping on craving at the next report, within the next 12 hours of the index report (Table 3). Neither coping effort during a specific stressful event (within individual) nor individual’s general likelihood of coping (between individual) prospectively impacted craving over the next 12 hours.

Table 3.

Trimmed Hierarchical Linear Model (HLM) Analysis of the Effects of Coping (Between t₀ and t₁) on Craving Over 12 Hours (t₁), Coping (Between t₀ and t₁) on Negative Affect (t₀), Negative Affect (t₀) on Craving Over 12 Hours (t₁)

Predictor	Coefficient	95% CI	Approx. df	p
Direct effect c path (12 h)
Smoking (since index to t1: Y/N)	.01	[−.28, .29]	259	.98
Coping (between t−1 and t0)
Within-subject coping	−.10	[−.38, .18]	259	.50
Between-subject coping	−.92	[−2.07, .23]	32	.11
a path (index report)
Smoking (since index to t1: Y/N)	−.07	[−.23, .09]	377	.37
Coping (between t−1 and t0)
Within-subject coping	−.29	[−.45, −.14]	377	<.001*
Between-subject coping	.32	[−.34, .99]	37	.33
b path (12 h)
Smoking (since index to t1: Y/N)	.03	[−.24, .32]	258	.79
Negative affect (index)
Within-subject negative affect	.22	[.04, .40]	258	.02*
Between-subject negative affect	.80	[.31, 1.29]	31	.002*
Coping (between t−1 and t0)
Within-subject coping	−.03	[−.31, .25]	258	.86
Between-subject coping	−1.14	[−2.16, −.11]	31	.03*

CI = confidence interval.

*p < .05

Coping and Negative Affect (a Path)

Higher within-subject coping predicted lower negative affect (p < .001), while there was no significant effect on negative affect for between-subject coping or smoking status (Tables 2 and 3). This suggests that whether an individual coped or not impacted their subsequent negative affect. However, individual differences in coping tendency did not predict negative affect.

Negative Affect and Craving (b Path)

The results showed that both higher within-subject negative affect (p < .001), and higher between-subject negative affect (p = .01) predicted higher craving (Table 2). That is, both an individual’s higher negative affect at the specific report and an individual’s higher general negative affect (ie, the average negative affect of all their EMA reports) predicted higher craving endorsed at the same report.

The results also showed that both higher within-subject negative affect (p = .02), and higher between-subject negative affect (p < .001) predicted higher prospective craving (Table 3), indicating that both the individual’s higher negative affect at the specific report and an individual’s higher general negative affect predicted higher craving over the next 12 hours.

Mediation Pathway

Our contemporaneous model showed that all paths in our hypotheses were supported. Within-subject coping significantly predicted lower craving (the c path; p = .02), and lower negative affect (the a path; p < .001). Additionally, reduced within-subject negative affect significantly predicted lower craving (the b path; p < .001). Furthermore, the b path model showed that the coefficient estimate of within-subject coping decreased (from −.28 to −.14) and became no longer significant (p = .22), indicating that the effect of coping on craving is mediated by negative affect.

The mediation pathway hypothesis was not supported for the prospective model, as there was no significant direct relationship between coping and craving in the c path (p = .50). However, we found that within-subject coping predicted lower negative affect (p < .001), supporting the a path. In the b path, we found that lower within-subject negative affect predicted lower subsequent craving (p = .02).

Discussion

The purpose of this study was to examine whether coping with stress reduces craving to smoke through reducing negative affect during an early quit attempt among smokers with mood disorders. Our results provided mixed support for our hypotheses in an ethnoculturally diverse sample. Stress coping predicted reduced subsequent craving through negative affect in the contemporaneous model, but not in the prospective model. Among smokers with mood disorders, the effect of stress coping on subsequent craving through negative affect appears to be short lived.

We observed that within-subject stress coping, but not between-subject coping, significantly predicted lower levels of subsequent craving, supporting our hypothesis on the total effect (c path) of coping on cravings. This indicates that coping with a stressful event may lead to lower craving, even after taking into account individual differences in the frequency of coping. This finding extends previous findings from an EMA study that stress coping significantly prevented lapses through reducing craving to smoke among smokers in a quit attempt.¹⁸ Our study is the first to demonstrate this effect among smokers with mental illness. While one study found that smokers with mental illness have high levels of avoidant coping through using substances and low levels of active coping,²⁵ our findings highlight the benefits of active coping among this vulnerable population. However, coping with stress did not impact craving in the prospective model (up to 12 hours following the report), suggesting that coping with stress may reduce craving for a limited period of time. Research has shown that cravings fluctuate rapidly and is influenced by a wide variety of factors (eg, cues, affect).^24,27,49 Perhaps craving may be more influenced by proximal factors of cues and affect, negating the previous effect of stress coping on craving.

Next, the effect of coping on negative affect (a path) in our hypothesis was supported; within-subject, but not between-subject coping significantly predicted subsequent negative affect. That is, coping with a stressful event may lower negative affect regardless of individual differences in the frequency of coping effort. Many smoking studies have demonstrated that coping with cravings reduces negative affect.^24,43 However, few studies have investigated the impact of coping with stress on negative affect among smokers. In an EMA study on daily adult smokers, coping with stress predicted reduced negative affect up to 4 hours later.¹⁹ Another daily diary study among college students also found that stress coping reduced negative affect.³⁸ However, this is the first study to demonstrate coping with specific stressful events effectively reduces negative affect during a quit attempt among smokers with mood disorders. This finding suggests that individuals heavily burdened with stress and negative affect benefit from coping with stress.

Consistent with our hypothesis on the effect of negative affect on craving (b path), negative affect (both within- and between-subject) was positively associated with both contemporaneous and prospective craving, controlling for coping. Those with higher average negative affect were more likely to endorse greater craving at any given report. Additionally, changes in negative affect within individual also predicted changes in craving, with the effect lasting up to 12 hours. The robust relationships between negative affect and craving are consistent with extant research.^5,30,50 For example, adult dependent smokers randomized to a negative mood induction condition in a laboratory study subsequently had greater craving compared with those in the neutral mood condition.⁵⁰ An EMA study among daily smokers found that even though the effect of negative affect on cravings decreased over time, it remained significant for up to 7 hours.⁴⁹ Thus, our findings suggest that negative affect may have an even longer effect on craving among smokers with mental illness.

In the contemporaneous model, all paths (ie, c, a, and b paths) were statistically significant in our hypothesized directions. Coping had a significant total effect on craving. When negative affect was included in the total effect (c path) model, the coefficient estimate for the direct effect of coping on craving (c′ path) was reduced and became nonsignificant, indicating that the relationship between coping and craving is mediated by negative affect. This finding is consistent with findings from an EMA study that coping with cravings predicted lower cravings in a quit attempt, explained by a reduction in negative affect.¹⁶ However, our study is the first EMA study to demonstrate that coping with stress is mediated by negative affect in the prediction of craving, as the above study measured coping through coping with cravings. Since previous research has demonstrated that individuals with mental illness have elevated stress and cravings, these findings are promising as they highlight the benefits of coping with stress in this vulnerable population.

Stress coping was not found to impact craving at the next report in the prospective model (within the next 12 hours), contrary to our hypothesis. However, the results from the prospective model showed that stress coping predicts lower subsequent negative affect (the a path), and reduced negative affect predicts lower craving for the next 12 hours (the b path). Thus, it is possible that coping may have indirectly affected craving through its impact on negative affect or through its impact on other factors discussed above (eg, smoking cues). Some argue there may be an indirect effect if both the a and b paths are significant, even without a significant c path.⁴⁴ Future research is needed to examine the indirect effects of coping with stress on craving.

Limitations

There are several limitations in this study. First, we only included reports with stressful events, and there were 10 participants excluded because they did not have any EMA reports documenting stressful events. While we did not find any demographic or baseline characteristic differences between the included and excluded participants, there may have been other differences. Next, there are potential biases if participants completed EMA reports during particular affective states or in certain contexts. Additionally, the individual differences in the number of completed reports with stress (median = 5, range = 75) could potentially lead to a higher likelihood of detecting within-subject differences from participants who completed more reports. However, researchers suggest that analyses using mixed models are suitable for unbalanced datasets.⁴² It is also difficult to infer complete mediation or indirect effects through the traditional mediation model, and many caution against using cross-sectional data to infer longitudinal mediation.^29,39 Negative affect was not manipulated, thus no causal link can be drawn from this study. Additionally, the time (M = approx. 5 hours) between the index and previous reports and the index and next reports may not be optimal to capture the effects of coping/negative affect on craving. In addition, given that the last smoking event could have occurred anytime between the last and current reports, using smoking status as a covariate in the models may not have effectively controlled for recent smoking. There may have been other covariates for which we did not account. However, we ran an analysis using the demographic variables as well as intervention condition as covariates and our findings did not change. Lastly, our study has a small sample size and had a high percentage of females, thus may not be generalizable to other clinical populations.

Conclusion

This study examined the impact of coping with stress on craving through its effect on negative affect in the context of a quit attempt among a diverse sample of adult smokers with mood disorders. The results suggest that among smokers with mood disorders, coping with stress may reduce craving through negative affect, but it may do so only within a short timeframe. This study was the first to use EMA to investigate the impact of coping with stress on craving to smoke among smokers with psychiatric disorders. The within-subject analyses of near real-time data allowed us to understand the impact of coping on subsequent negative affect and craving within an individual. Our study had a strength of having a sample that is hard to reach and ethnoculturally diverse. Future research could utilize an indirect effects model to assess the impact of stress coping and negative affect on craving. Since stress coping impacts craving within a limited timeframe, additional research is needed to identify the specific length of time that this effect lasts among smokers with mental illness. The finding that coping is beneficial in reducing negative affect and cravings could guide research on helping smokers with mental illness cope when experiencing stress.

Supplementary Material

A Contributorship Form detailing each author’s specific involvement with this content, as well as any supplementary data, are available online at https://academic.oup.com/ntr.

Funding

The project described is supported by Award Number NIDA grant R34 DA037364 from the National Institute on Drug Abuse (PIs: Haruka Minami, PhD and Richard A. Brown, PhD).

Declaration of Interests

Dr Nahvi receives investigator-initiated grant support from Pfizer.

Ethical Approval

The authors assert that all procedures contributing to this work comply with the ethical standards of the relevant national and institutional committees on human experimentation and with the Helsinki Declaration of 1975, as revised in 2008.

Data Availability

Data will be available on request.