Cigarette Smoking and Depression Among U.S. Veterans: Longitudinal Associations With Posttraumatic Stress Disorder

Abstract

Introduction

Cigarette smoking and depression are associated with morbidity and mortality. Among veterans, approximately 22% are current smokers and 11%–15% have been diagnosed with depression. Although prior research suggests a strong association between smoking and depression among veterans, little research has examined trajectories of smoking and depressive symptoms and their correlates over time in this population.

Aims and Methods

Using parallel process growth curve modeling, we examined the longitudinal relationship between smoking and depression and tested whether posttraumatic stress disorder (PTSD) symptoms predict smoking and depression trajectories over 18 months (February 2020–August 2021). Veterans were recruited for an online, longitudinal study and responded to surveys across five-time points (baseline N = 1230; retention = 79.3%–83.3% across waves).

Results

Associations indicated that more frequent smoking at baseline was associated with steeper increases in depression symptom severity, and greater depression severity at baseline was associated with a less steep decrease in smoking frequency over time. PTSD was associated with less smoking at time 1 but more frequent smoking at times 3–5 as well as greater depression across all time points.

Conclusions

Findings provide support that the growth trajectories of smoking and depression are linked, and PTSD symptoms are associated with these trajectories among veterans. Addressing these factors simultaneously in veteran treatment centers or through tobacco cessation efforts may be beneficial.

Implications

This study offers strong evidence that the growth trajectories of smoking and depression are linked, and PTSD symptoms affect these trajectories among veterans, who represent a largely understudied population despite high rates of substance use and mental health problems. Results of this study strengthen the case for a more integrated treatment approach in which both smoking and mental health concerns are simultaneously addressed, which may yield more beneficial physical health and clinical outcomes for post-9/11 veterans.

Introduction

Compared to individuals without depression, individuals living with depression are more likely to smoke cigarettes.^1–3 Alternatively, among those who smoke cigarettes, those with cooccurring depression smoke more cigarettes per day, experience worse nicotine withdrawal symptoms, and are less likely to successfully quit smoking.^1–3 American veterans are an especially vulnerable population, with one in five veterans reporting past month smoking compared to one in seven among civilian samples.⁴ Rates of smoking appear even higher among Operation Enduring Freedom/Operation Iraqi Freedom (OEF/OIF; often also called “post 9/11 veterans”) veterans with estimates of one in three reporting past month smoking.^5,6 In addition, the consequences of these higher smoking rates are compounded by the high prevalence of depression in this population.^7,8 In fact, rates of current smoking are nearly one in two among OEF/IEF veterans with a psychiatric disorder.⁹

The incentive learning account is a theoretical model that posits that smokers with depression tend to experience more negative affective states brought on by nicotine withdrawal and place greater reinforcement value toward smoking during withdrawal states.¹⁰ Over time, smokers with depression may fail to distinguish between negative affect brought on by withdrawal and negative affect provoked by other triggers (eg, stress), and thus, continuously turn to smoking in response to both states regardless of the source or trigger of the negative affect. In other words, depression symptoms may trigger smoking behaviors during attempts to alleviate oneself from the negative affect associated with experiencing depression. In turn, nicotine withdrawal may trigger additional negative affective states, which may then trigger additional smoking behaviors. To test whether this framework is applicable among veterans, a population with high rates of both depression and smoking, longitudinal studies are needed that examine smoking and depression symptoms simultaneously over time. Such work may help in the development of more intensive smoking cessation efforts in conjunction with depression treatment. In fact, there are currently no known psychotherapeutic or behavioral interventions specifically for veterans that concurrently target both smoking cessation and depression needs.

One additional factor known to trigger negative affective states and is associated with increased smoking and depression among veterans is posttraumatic stress disorder (PTSD). Like the incentive learning account’s position that depression and smoking may be mutually maintaining, the mutual maintenance model takes into account the bidirectional or mutually maintaining relationship between PTSD and physical symptoms.¹¹ In line with the mutual maintenance model, PTSD symptoms may trigger smoking behaviors. In turn, nicotine withdrawal symptoms (eg, physical symptoms such as difficulty sleeping or feeling restless) may then exacerbate PTSD symptoms, potentially contributing to increased smoking behaviors. Previous longitudinal studies of veterans have shown a strong, positive association between PTSD symptom severity and smoking, including smoking initiation,^12,13 as well as PTSD and major depressive disorder (MDD) symptom severity.¹⁴ PTSD and depression are among the most prevalent mental health problems among veterans¹⁵ and often cooccur in as many as half of veteran samples.¹⁶ Furthermore, PTSD is associated with a two-fold higher risk of current smoking among post-9/11 veterans,¹⁷ and veterans who experience reductions in PTSD symptoms are more likely to quit smoking.¹⁸ However, there is a paucity of longitudinal research that examines whether PTSD symptoms impact the trajectories of smoking and depression among veterans over time. Such research may provide key information in the design of treatments for veterans with cooccurring smoking cessation and depression needs.

Stay Quit Coach is a smartphone-based application and is the only known smoking cessation intervention designed for veterans with PTSD, with 1 small trial of 11 smokers reporting that only one participant was abstinent at the end of treatment.¹⁹ Furthermore, reducing PTSD symptoms was not a target of the trial study, as achieving prolonged abstinence was the study’s main objective. Whether the intervention led to reductions in smoking, as opposed to complete abstinence, was also not reported. Another study examined the effects of Stay Quit Coach as a supplement to 8 weekly sessions of an integrated care smoking cessation treatment program and reported that 6 out of 17 participants (35.3%) who completed all sessions reported abstinence at the 3-month follow-up.²⁰ There were no significant reductions in PTSD symptoms and results should be interpreted with caution given the lack of a control group. Continued research may help identify whether veterans who smoke, experience depression, and have PTSD may also benefit from a more integrated form of treatment where all three factors (smoking, depression, and PTSD) are addressed.

The Current Study

Despite the strong association between cigarette smoking and depression, to date, there are few studies that examine longitudinal trajectories of smoking and depression among veterans and how PTSD symptoms influence these trajectories. Building on previous work and theory, we sought to understand the longitudinal relationship between smoking, depression, and PTSD among a large sample of post-9/11 veterans. Doing so may help identify unique intervention targets among a population with incredibly high smoking, depression, and PTSD rates. The first aim is to assess simultaneous changes in cigarette smoking and depression symptoms using parallel process growth curve modeling. Doing this will allow us to explore, for example, how change in cigarette smoking is associated with change in depression. The second aim is to explore how PTSD symptoms influence parallel and emergent trajectories in cigarette smoking and depression symptoms over time.

Methods

Participants and Procedures

Veterans between the ages of 18 and 40 who had separated from the Air Force, Army, Marine Corps, and Navy were eligible and recruited in February 2020 as part of a larger study about the behavioral and mental health of young adult veterans during the coronavirus disease 2019 (COVID-19) pandemic. Advertisements were distributed on general and military-specific social media platforms (ie, Facebook, Instagram, RallyPoint, and We Are the Mighty). Eligible veterans who consented to participate in the study completed a 30-minute online survey and received a $20 Amazon gift card for their participation. Internal validation checks were conducted to remove participants who provided potentially falsified data to receive incentives, such as endorsing inconsistent responses between items within and across surveys (eg, branch, rank, and pay grade matches), completing the survey too quickly or selecting the same response option throughout, and attempting to access the survey multiple times by reviewing IP addresses. This series of validation checks removed 625 individuals from the sample. The final baseline sample consisted of 1230 veterans.

Participants were later invited to complete four follow-up electronic surveys over the course of the pandemic. In order to minimize the number of participants that may potentially be lost to follow-up, reminder emails were also provided to those who did not complete the follow-up surveys after a certain number of days. These follow-up surveys were sent via e-mail at 6 months (time 2; N = 1025; 83.3% retention from baseline), 9 months (time 3; N = 1006; 81.8% retention from baseline), 12 months (time 4; N = 1005; 81.7% retention from baseline), and 18 months (time 5; N = 976; 79.3% retention from baseline) after completion of the initial survey (time 1). Participants received an Amazon gift card for participating in each survey: $30 at time 2, $40 at time 3, $50 at time 4, and $30 at time 5. All study materials and procedures were approved by the local Institutional Review Board. See Davis et al. (2021)²¹ for more information on participant recruitment.

Measures

Demographic and Military Characteristics

Variables controlled for in analyses were sex (male and female) and race and ethnicity (due to small group sizes for racial/ethnic minority groups, racial and ethnic categories were dichotomized to white and nonwhite to facilitate analyses) based on prior research.^22,23

Main outcomes

Cigarette Smoking

Participants were asked to report the number of days they smoked cigarettes within the last 30 days. Scores ranged from 0 to 30, with higher values indicating greater number of days smoked. Because this was a secondary analysis of a study on alcohol use outcomes for veterans, quantity of cigarettes smoked was not assessed.

Depression

Symptoms of depression were measured with the Patient Health Questionnaire 8-item (PHQ-8).²⁴ Symptoms experienced over the past 2 weeks (eg, feeling down, hopeless, or depressed) were rated from not at all (0) to nearly every day (3). The responses are summed with scores ranging from 0 to 24, with higher values representing more severe depression symptoms. This measure had a mean reliability estimate of α = 0.71 at all five-time points for this sample. Because suicidality was not a focus of the parent study, the PHQ-8 was utilized over the PHQ-9 to measure depression symptoms. The PHQ-8 is considered to be a reliable and valid measure with Cronbach’s alphas of 0.82–0.89 in other samples,^25,26 has become increasingly used in research, and demonstrates high sensitivity and specificity to the PHQ-9.²⁷

PTSD Symptom Severity

PTSD symptom severity was treated as a time-varying covariate. It was assessed using the Posttraumatic Stress Disorder Checklist for DSM-5 (PCL-5)²⁸ which consists of 20 items on a five-point Likert scale (0 = not at all to 4 = extremely). Participants self-reported how often they experienced 20 PTSD symptoms (eg, avoidance of internal or external reminders of the traumatic experience, feeling jumpy or easily startled) in the past month. Responses were summed with total scores ranging from 0 to 80. The PCL-5 had a mean reliability estimate of α = 0.90 across the five-time points in the current sample. This measure is widely considered to be a reliable and valid measure with Cronbach’s alphas of 0.94–0.96 in other samples.^28,29

Statistical Analyses

See Figure 1 for a conceptual diagram of our model. To address aim 1 (assess simultaneous changes in cigarette smoking and depression), we estimated a taxonomy of latent growth curve models.³⁰ First, we fit an unconditional latent growth model to determine the functional form of the data for both depression and cigarette smoking, separately. For each model, we tested for random intercepts and random linear or quadratic slopes (vs. fixed). Given the non-linear trends of both cigarette smoking and depression, we chose to employ latent basis growth modeling, which is a flexible version of growth curve modeling.³¹ Here, the slopes are not constrained to linear time but are instead allowed to be freely estimated. This is an ideal approach, as it allows us to determine the optimal growth trajectories from the observed data (the result of these growth models fits the functional form of the data).

Figure 1.

Conceptual parallel process growth model with time-varying covariate. The time-varying covariate depicted here is PTSD symptom severity. Time-invariant covariates are a representation of both sex and race or ethnicity.

Next, we introduced a parallel process latent growth model. Here, smoking and depression latent growth models were estimated simultaneously. Parallel process latent growth models allow us to examine trajectories of change (ie, growth processes) in two or more variables simultaneously, making it possible to examine how the intercept and slope in one variable are associated with the intercept and slope in the other. In the present model, we used a directional specification. Here, the intercepts and slopes of both depression and cigarette use are correlated. However, when specifying a directional model, regression is used to understand the effect of how intercepts for one factor (eg, depression starting point) influence change in the second factor (eg, cigarette change).

To address aim 2 (assess the effect of our time-varying covariate), we introduced PTSD symptomology as a time-varying covariate. To do this, at each time point, we regressed PTSD symptomology onto the contemporaneously observed outcome variables for smoking and depression. This allowed us to determine the effect of PTSD symptoms on our smoking and depression outcomes over time, above and beyond the effects of the underlying parallel process growth model. We first introduced PTSD as constrained effect (eg, effects are constrained to be equal over time) and tested this model against one in which PTSD was unconstrained (eg, freely estimated at each time point). We used the difference in negative two log-likelihood ratio test to determine if constrained versus freely estimated model fits the data best.³² The final model included PTSD symptoms as a constrained or unconstrained effect, based on nested model equality constraint test. Doing this also allowed us to determine if PTSD had consistent, stronger, or weaker influences over time. We used traditional fit indices including, root mean square error of approximation (RMSEA), comparative fit index (CFI), and standardized root mean square residual (SRMR) to assess goodness of model fit. All models were estimated in Mplus 8.6,³³ which uses full information maximum likelihood to aid in missing data analysis for outcome variables, with latent growth modeling retaining in analyses all participants who completed baseline measures. All models included sex and race/ethnicity as covariates on intercepts and slopes for both cigarette smoking and depression.

Results

Sample Characteristics

On average, participants were 34.5 years old, and 88.7% identified as male gender. Veterans served in the military for an average of 9.6 years (SD = 3.7) and had been out of the military for an average of 5 years (SD = 3.5). See Table 1 for additional information on participant demographics.

Table 1.

Sample Characteristics

Variable	n (%) or M (SD)
Sex
Male	1091 (88.7%)
Female	138 (11.2%)
Race and Ethnicity
White	975 (79.3%)
Nonwhite	255 (20.7%)
Baseline Posttraumatic Stress Disorder Symptoms	22.0 (15.8)
Days of past 30-day cigarette smoking
Time 1	14.0 (10.0)
Time 2	9.0 (10.1)
Time 3	3.4 (8.6)
Time 4	3.6 (8.9)
Time 5	4.3 (9.8)
Depression symptoms
Time 1	7.7 (4.9)
Time 2	7.2 (4.3)
Time 3	8.8 (3.7)
Time 4	8.7 (3.8)
Time 5	8.4 (3.6)

M = mean; SD = standard deviation.

Cigarette Smoking and Depression Symptoms

Prior to fitting parallel process models, we estimated latent growth models for depression and cigarette smoking, separately to determine the functional form of the data for both depression and cigarette smoking. Veterans, on average, started with a baseline score of 7.76 for depression (intercept) with significant increases in depression over time (b = 1.81, p < .001). Intercept values for cigarette smoking indicated veterans had an average of 14 days of cigarette use in the past 30 days and showed a significant decrease in cigarette use (b = −17.74, p < .001) over the study period.

Next, to address aim 1, we fit a parallel process growth model (ie, two latent growth models are developed simultaneously and the growth factors [intercepts and slopes for depression and smoking] are allowed to covary). This originally proposed model did not fit the data well: Model χ² = 528.43, p < .001; CFI = 0.86; TLI: 0.85; RMSEA = 0.10, 90% CI [0.09, 0.10]; SRMR = 0.08; thus we used modification indices to inform our final model.³⁴ This model fit the data well: Model χ² = 259.27, p < .001; CFI = 0.93; TLI: 0.92; RMSEA = 0.07, 90% CI [0.06, 0.08]; SRMR = 0.07. The final parameter estimates are presented at the top of Table 2. Cross-construct associations (associations between intercepts and slopes) indicate that more frequent cigarette smoking at baseline is associated with steeper increases in depression symptom severity (b = 0.27), whereas a higher baseline level of depression is associated with a less steep decrease in smoking frequency (b = 0.63). Furthermore, higher baseline frequency of smoking is associated with steeper decreases in smoking frequency (b = −0.40) and higher baseline level of depression is associated with a less steep increase in depression (b = −0.59). See Supplementary Table 1 for results with time-invariant control variables.

Table 2.

Parallel Process Model for Cigarette Smoking and Depression With Time-Varying Covariates

	Parameters	SE	p-value
Cigarette smoking model
Growth parameters
Intercept	14.58	1.17	<.001
Slope	−17.74	1.64	<.001
Variance
Intercept	55.74	8.20	<.001
Slope	111.07	11.03	<.001
Depression model
Growth parameters
Intercept	0.77	0.28	.005
Slope	1.81	0.33	<.001
Variance
Intercept	11.76	0.75	<.001
Slope	2.28	0.52	<.001
Parallel process model
Intercept cigarette with slope depression	0.27	0.06	<.001
Intercept cigarette with slope cigarette	−0.40	0.07	<.001
Intercept depression with slope cigarette	0.63	0.07	<.001
Intercept depression and slope depression	−0.59	0.05	<.001
Cigarette smoking model
PTSD (time 1)	−0.30	0.03	<.001
PTSD (time 2)	−0.02	0.02	.480
PTSD (time 3)	0.11	0.03	<.001
PTSD (time 4)	0.13	0.03	<.001
PTSD (time 5)	0.21	0.03	<.001
Depression model
PTSD (time 1)	0.85	0.01	<.001
PTSD (time 2)	0.74	0.01	<.001
PTSD (time 3)	0.70	0.02	<.001
PTSD (time 4)	0.67	0.02	<.001
PTSD (time 5)	0.66	0.02	<.001

SE = standard error; PTSD = posttraumatic stress disorder.

Time-Varying Covariate: PTSD Symptoms

To address aim 2, we introduced PTSD symptom severity as our time-varying covariate into the model. This model fit the data well: Model χ² = 438.69, p < .001; CFI = 0.94; TLI: 0.93; RMSEA = 0.06, 90% CI [0.06, 0.07]; SRMR = 0.09. In our final model, looking at our time-varying covariate predicting cigarette smoking, PTSD was allowed to be freely estimated (based on the difference in negative two log-likelihood ratio test) and resulted in less frequent cigarette smoking at time 1 but more frequent smoking at times 3–5. This increase in smoking from time 1 to time 5 was significantly different (Wald test of parameter constraints value = 185.95 p < .001), demonstrating that PTSD had an accelerating effect on cigarette smoking over time. PTSD was not associated with cigarette smoking at time 2. PTSD was also allowed to be freely estimated for the depression portion of the model. Here, PTSD was associated with greater, contemporaneous depression symptoms at each time point. However, though these effects were positive, the valence got smaller over time. Results of model tests between time 1 and time 5 indicated effects were significantly different (Wald test of parameter constraints value = 89.63, p < .001), demonstrating that PTSD had a decelerating effect on depression over time.

Discussion

Veterans report relatively high rates of both current smoking (21.6%)³⁵ and depression (11.4%–15.0%).³⁶ Because smoking and depression are both leading causes of disability, morbidity, and mortality,^37,38 a primary focus of the study was to examine simultaneous changes in smoking and depression among veterans over the course of 18 months. Individual trajectories of depression were associated with rates of change in smoking frequency and individual trajectories of smoking frequency were associated with rates of change in depression. This study used a novel approach to rigorously integrate research on two of some of the most critical behavioral health problems faced by veterans, smoking and depression, and demonstrate a dynamic relationship between them.

Data were collected directly prior to and during the COVID-19 pandemic, which offered a unique opportunity to examine substance use and mental health symptoms during a volatile period for many Americans as they adapted to stay-at-home orders and subsequent social restrictions. Consistent with previous research showing widespread decreases in nicotine use during the pandemic among the general population,^39,40 which could be attributed to perceptions that smoking would increase the risk of contracting the virus or having a more severe case,⁴¹ cigarette smoking frequency in this sample also decreased over the course of the study. However, there was a slower decline in smoking frequency among those with greater baseline depression symptoms. The findings that baseline depression is associated with the rate of smoking frequency are in line with the incentive learning account’s assertion that depression primes smoking behavior,¹⁰ suggests that targeting depression during the onset of smoking cessation treatment may aid veterans to achieve successful smoking abstinence more quickly. In accordance with this theoretical framework, treatment may need to focus on correcting adverse depressive states and address beliefs about the high value of smoking when experiencing these negative states. However, additional research is needed to determine whether these would improve cessation outcomes in depressed smokers.

Further supporting the incentive learning account, a higher starting value in smoking frequency was associated with a steeper increase in depression symptoms over time, suggesting that focusing on reducing smoking may help prevent worsening depression symptomology over time. This is further supported by previous work showing that smoking actually contributes to worsening mental health symptoms over time and that quitting smoking leads to improved mental health.⁴² In line with the incentive learning theory, smokers over time develop an expectation that greater smoking is associated with greater relief from aversive depressive states (ie, greater negative reinforcement value).¹⁰ As depression is associated with more severe withdrawal from nicotine, this then increases the reinforcement value of engaging in smoking among depressed smokers. This may help explain our finding that higher baseline values in smoking frequency were associated with a steeper increase in depression symptoms over the study period.

Studies with civilian samples have yielded similar findings, showing that worsening mental health was linked to increased smoking during the pandemic.^43,44 However, these studies were limited by their cross-sectional design. It is important to note that the present study’s findings are based on data from the overall sample. Another study examined the role that affects, a component of depression, has on smoking behaviors and specifically examined sex differences in these associations. In this examination of affect-based triggers of smoking during a cessation attempt, men reported higher levels of positive affect-induced lapses compared to women.⁴⁵ There were no overall sex differences in negative affect-induced smoking; however, while levels of negative affect-induced smoking decreased over time among men, these levels persisted among women. It is possible that similar sex differences are present in the association between the rate of change in depression and smoking over time. Future studies may focus on larger recruitment of female veterans in order to test for such differences.

We also tested whether PTSD symptoms affected changes in smoking and depression in this sample. Understanding factors that are associated with smoking and depression over time may yield useful clinical and health-related implications that aid prevention and intervention/cessation efforts. In line with the mutual maintenance model, PTSD demonstrated a positive and steadily increasing effect on smoking frequency across the study period. This mutual maintenance model has become widely accepted even though few longitudinal studies of the potential bidirectional relationship between PTSD and physical symptoms exist.⁴⁶ Of note, the current study did not examine physical symptoms, such as nicotine withdrawal; thus, future research should consider the role of withdrawal in the longitudinal relationships between PTSD, smoking, and depression. Nevertheless, our results are consistent with prior longitudinal work demonstrating positive associations between PTSD and smoking among post-9/11 veterans¹² and represent one of the few veteran-specific studies to report these longitudinal associations.^47,48 Findings suggest that addressing PTSD symptoms in smoking cessation interventions or promoting tobacco cessation and relapse prevention skills within existing evidence-based treatments for PTSD may be important to prioritize among clinicians and other health professionals who provide care to veterans. In fact, there is evidence that smoking cessation combined with PTSD treatment may be effective in facilitating prolonged abstinence from tobacco use among veterans.⁴⁹

PTSD was also associated with greater depression symptom severity at all five-time points. PTSD and depression are among the most prevalent mental health problems faced by military veterans.¹⁵ Individuals with comorbid PTSD and MDD experience worse outcomes than those with either condition alone, including poorer social and occupational functioning⁵⁰ and persistent PTSD symptoms and suicidality.^51,52 Because PTSD and MDD are highly comorbid among military service members^53,54 and this comorbidity is associated with greater negative sequela for individuals with both disorders, continued efforts to improve outcomes for these individuals, such as efforts to increase support and clinical outreach, are essential.

Strengths and Limitations

There are limitations to this study. First, we reported secondary findings from a larger parent study that focused on addressing heavy alcohol use among non-treatment-seeking young adult veterans (eg, veterans outside of Veterans Affairs [VA]

settings). Thus, conclusions may not be generalizable to veterans who are older, endorse lower alcohol use rates, or are receiving services from the VA. Second, our sample consisted of primarily white veterans. Behavioral health researchers who focus on veteran populations should expand upon our work through recruitment of a more ethnically/racially diverse sample. Furthermore, because this study utilized secondary data from a preexisting study that primarily focused on alcohol use outcomes, information about the average number of cigarettes smoked per day by participants were not gathered and thus, cigarette frequency, and not quantity on use days in the last 30 days, was solely assessed. Including a measure for quantity of smoking would have made testing for differences between lighter quantity of use (eg, smoking one cigarette per day) and heavier quantity of use (eg, smoking 20 cigarettes per day) possible and provided useful additional insight. Future replication studies may benefit from including quantity and frequency indices as variables of interest. Finally, data were collected during the COVID-19 pandemic during which cigarette smoking frequency was decreasing, limiting our ability to generalize results outside of this period. Nonetheless, this can serve as a strength of the study, as the COVID-19 pandemic marked a particularly stressful and risky time for veterans (and the general population) where fluctuations in substance use behaviors and mental health symptoms were common.⁵⁵ Thus, an examination of the trajectories of substance use behaviors during the pandemic is timely and relevant. Relatedly, in contrast to the majority of other behavioral health studies during the COVID-19 pandemic that relied on cross-sectional study designs,⁵⁶ this study utilized data collected longitudinally directly prior to and during the pandemic for a duration of 18 months. The study has several other key strengths including the achievement of high retention rates. In contrast to other studies with veterans that prohibit incentives,⁵⁷ it is possible that providing gift card monetary incentives is effective for retention. Lastly, the study utilized a large sample of veterans, who represent a largely understudied group that is well known to experience high rates of mental health problems and tobacco use. Hence, this study offers additional insights that may help address the incredible health and economic burden of tobacco use and depression for military veterans.

Clinical Implications

From a behavioral health perspective, these results offer compelling evidence that the growth trajectories of smoking and depression are linked, and PTSD symptoms affect these trajectories. Inaccurate concerns that treating smoking among individuals with mental health problems would worsen depression, other mental health symptoms, and other substance use have been significant barriers to smoking cessation efforts in mental health settings.⁵⁸ The results of this study strengthen the case for a more integrated treatment approach, such as screening smokers or veterans with depression for concerns related to PTSD, and providing additional support as needed. For example, smokers in cessation programs who also screen positive for depression or PTSD may benefit from receiving additional treatment for depression and PTSD, as these may interfere with cessation goals. In fact, in a study with veterans quitting smoking, smoking abstinence at follow-up was associated with improvements in mental health, including depression and emotional lability.⁵⁹ Tobacco cessation counseling (ie, individual, group, telephone, or video counseling) and medications, such as bupropion (a medication approved by the Food and Drug Administration for the treatment of both depression and smoking), are offered at all VA medical centers; thus, veterans who smoke should be encouraged to inquire about and utilize these available resources. Integrated efforts such as these in which both behavioral and mental health concerns are simultaneously targeted may be impactful.

Conclusion

Smoking and depression are both associated with profound negative outcomes; thus, understanding the association between them and factors that contribute to greater smoking and depression symptoms among veterans is paramount. In this large sample of post-9/11 veterans, veterans who smoked more frequently prior to the start of the pandemic reported steeper increases in depression symptoms over the course of the pandemic. On the other hand, veterans who reported greater depression symptoms prior to the start of the pandemic reported a slower decline in their smoking frequency during the pandemic. PTSD symptoms were positively associated with cigarette smoking and this effect increased (accelerated) over time. Future studies should include an examination of other factors that may influence smoking and depression trajectories among veterans and/or strive to replicate these findings outside of the pandemic era. It may also be essential to develop interventions that target smoking, depression, and PTSD simultaneously for veterans.

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

This research was funded by grant R01AA026575 from the National Institute on Alcohol Abuse and Alcoholism (NIAAA), supplement R01AA026575-02S1, and a Keck School of Medicine COVID-19 Research Funding Grant awarded to ERP.

Declaration of Interests

All authors declare no conflict of interests.

Data Availability

We will make the dataset and associated documentation available outside our team under a data- sharing agreement that provides for (1) a commitment to using the data only for research purposes and not to identify any individual participant, (2) a commitment to securing the data using appropriate computer technology, and (3) a commitment to destroying or returning the data after analyses are completed. Contact the senior author if interested in obtaining data from this project.