Pain Status as a Predictor of Smoking Cessation Initiation, Lapse, and Relapse

Abstract

Introduction

Pain and cigarette smoking are highly prevalent and frequently co-occurring conditions that interact in the manner of a positive feedback loop. Despite initial evidence that smokers with co-occurring pain may experience greater difficulty quitting, we are unaware of previous research that has tested prospective associations between pain status and the attainment of smoking cessation milestones.

Aims and Methods

This study examined past 2-week pain status as a predictor of cessation milestones among current smokers who were motivated to quit (Sample 1; N = 301) and smokers who recently initiated a cessation attempt (Sample 2; N = 242). Cessation milestones included initiation of a quit attempt and 7-day point prevalence abstinence (PPA; Sample 1), lapse/relapse (Sample 2), and 7-day PPA at 2-month follow-up (both samples). Indirect associations between pain status and cessation milestones via confidence in quitting and nicotine withdrawal were also examined.

Results

Smokers with pain (vs. no pain) were as follows: less likely to initiate a quit attempt and achieve 7-day PPA; more likely to lapse and/or relapse; and less likely to report 7-day PPA at follow-up. Pain status was indirectly associated with latency cessation milestones via confidence in quitting and with latency to lapse via withdrawal severity.

Conclusions

This study demonstrated that pain status can predict smoking cessation outcomes. Clinical implications include the need to assess pain in the context of quitting and that smokers with co-occurring pain may benefit from tailored/integrated cessation interventions.

Implications

A growing empirical literature indicates that the presence of co-occurring pain probably contributes to the maintenance of cigarette dependence. The current results provide novel evidence that smokers with co-occurring past 2-week pain are less likely to initiate a quit attempt and maintain smoking abstinence than smokers without co-occurring pain. These findings suggest that smokers with pain face unique barriers to quitting and underscore the utility of assessing and addressing pain among all smokers who are planning a smoking cessation attempt.

Introduction

Pain and tobacco cigarette smoking are highly prevalent and frequently co-occurring conditions that produce a synergistic burden on individuals, healthcare systems, and the U.S. economy (e.g., medical and lost productivity costs estimated to exceed $900 billion, annually).^1–4 Whereas the prevalence of cigarette smoking in the general population has decreased to approximately 14% over the past 15–20 years,^5,6 rates of smoking among treatment-seeking pain patients have remained steady.⁷ Indeed, nationally representative survey data indicate that individuals with chronic pain (vs. no chronic pain) are two times more likely to currently smoke cigarettes,⁸ and the prevalence of current smoking among treatment-seeking pain patients has been estimated to range from 24% to 68%.^7,9

An established reciprocal model posits that pain and cigarette smoking interact in the manner of a positive feedback loop, resulting in greater pain and the maintenance of tobacco dependence.^1,2,10 Consistent with this perspective, regular smoking has been implicated in the onset and severity of painful conditions,^11,12 and situational pain has been shown to motivate smoking urge and behavior.^13,14 The presence of co-occurring pain has also been associated with lower cessation self-efficacy,^15,16 and expectations for more severe nicotine withdrawal during future quit attempts.¹⁶ Both cessation self-efficacy and nicotine withdrawal severity have been identified as central precipitants of relapse among nonpain samples.^17–20 Indeed, pain may increase the risk for experiencing difficulties with initiating and maintaining cessation, via associations with confidence in quitting and nicotine withdrawal.^17–20 However, despite evidence of covariation between pain status and cessation-relevant cognitions, and suggestions that smokers with co-occurring pain probably constitute a recalcitrant subgroup that faces unique challenges to quitting,^{2,10,15,21,22} surprisingly little research has examined the role of pain in the initiation and maintenance of smoking cessation.

In terms of initiation, one electronic daily diary study showed that patients undergoing treatment for cancer (N = 34) who reported greater pain during the first 2 weeks of smoking cessation treatment were less likely to make a quit attempt during the study period.²³ In terms of cessation, greater pain-related anxiety (i.e., tendency to respond to pain with anxiety or fear) and sensitivity to experimental pain induction have each been associated with an increased likelihood of early relapse following a quit attempt.^24,25 In addition, one study of people living with HIV/AIDS (N = 474) revealed that lower self-reported pain intensity over the course of a 3-month cell phone-delivered smoking cessation intervention was associated with greater likelihood of 24-hour and 7-day point prevalence abstinence from treatment end through 12-month follow-up.²⁶ Although these studies provide initial evidence that pain-relevant factors can uniquely influence pre-cessation processes and post-quit outcomes, generalizability is inherently limited, and we are unaware of previous research that has tested prospective associations between pain status and cessation milestones among smokers recruited from the general population. Given that much of the extant research examining pain in relation to smoking cessation-relevant constructs/processes has been retrospective or cross-sectional in nature, additional prospective analysis is sorely needed to establish predictive utility and clarify the temporal sequence of pain status in the prediction of cessation milestones.²⁷

The primary goal of the current longitudinal analysis was to examine past 2-week pain status as a predictor of cessation milestones across two different samples of smokers, including those who were motivated to quit but had not yet initiated a cessation attempt (Sample 1: Current Smokers) and those who had initiated a cessation attempt in the past 5 weeks (Sample 2: Recent Quitters). For Sample 1, we hypothesized that current smokers with co-occurring pain (vs. no co-occurring pain) would be less likely to initiate a quit attempt during the study period and report 7-day point prevalence abstinence (PPA) at 2-month follow-up. For Sample 2, we hypothesized that recent quitters with co-occurring pain (vs. no co-occurring pain) would be more likely to lapse and relapse during the study period and less likely to report 7-day PPA at 2-month follow-up. A secondary goal of this study was to examine potential mechanisms underlying the hypothesized relations between pain and cessation milestones. Specifically, we hypothesized that positive past 2-week pain status would be associated with lower confidence in quitting/remaining quit (among both samples) and greater nicotine withdrawal severity (among Sample 2) at baseline and that pain status would be indirectly associated with cessation milestones (i.e., latency to initiate a quit attempt, achieve 7-day PPA, lapse/relapse) via confidence in quitting and nicotine withdrawal severity.

Materials and Methods

Participants

Both samples were recruited using nationwide ads posted on Craigslist and ResearchMatch as part of a primary study examining smoking cessation fatigue as a predictor of cessation milestones.²⁸ Sample 1 consisted of N = 301 current tobacco cigarette smokers (≥10 cigarettes per day on at least 25 of the last 30 days) who reported strong motivation to quit smoking within the next 30 days (≥7 on a 0–10 scale). Sample 2 consisted of N = 242 recent quitters who initiated a smoking cessation attempt within the past 5 weeks, reported continued abstinence at the baseline assessment, and reported smoking at least 10 cigarettes per day on at least 25 days per month prior to their quit date. Participants in both samples were required to be between 18 and 65 years of age and speak English.

Procedure

After completing an online screening survey, eligible participants completed baseline measures. Participants were then recontacted via email every 2 weeks (five assessments in total) over an 8-week follow-up period. All data were collected using REDCap electronic data capture. Following completion of study assessments, participants were provided with contact information for a toll-free national smoking cessation quitline.

Measures

Sociodemographic and Smoking Characteristics

A range of sociodemographic (e.g., age, gender, race/ethnicity) and smoking characteristics were assessed via self-report at the baseline assessment. Cigarette dependence was measured using the Heaviness of Smoking Index, which is comprised of two items (i.e., “How soon after you wake up do you smoke your first cigarette?” and “How many cigarettes per day do you smoke?”).²⁹ Responses were summed to generate a total score (range: 0–6). Heaviness of Smoking Index scores have previously been shown to predict smoking abstinence outcomes.³⁰

Pain Status

The presence of co-occurring pain was assessed at baseline using a single binary item: “Have you had any significant pain in the past two weeks?” The presence of past 2-week pain has been positively correlated with more comprehensive measures of pain status/severity (e.g., Graded Chronic Pain Scale),³¹ and previous work has demonstrated clinically relevant associations between past 1- and 2-week pain status and cigarettes smoked per day/tobacco dependence scores.^23,32

Confidence in Quitting Smoking/Remaining Abstinent

Confidence in quitting or staying quit over the next 30 days was assessed using a single item at baseline (“How confident are you that you can quit smoking cigarettes or stay quit over the next 30 days?”). Response options ranged from 0 (not at all confident) to 10 (extremely confident).³³ Similar Likert-type measures assessing confidence in quitting smoking have consistently been associated with smoking cessation outcomes.^34,35

Nicotine Withdrawal Severity

The Minnesota Nicotine Withdrawal Scale was used to assess the severity of nine prototypical nicotine withdrawal symptoms over the past 24 hours (e.g., desire or craving to smoke) on a scale ranging from 0 (none) to 4 (severe).³⁶ Items were averaged to generate a total withdrawal severity score. Recent quitters (Sample 2) completed the Minnesota Nicotine Withdrawal Scale at baseline.

Cessation Milestones

Cessation milestones were assessed at each follow-up time point (i.e., weeks 2, 4, 6, and 8). Specifically, participants were asked to indicate whether they had smoked any cigarettes (even a puff) since the last assessment. This single-item approach has shown strong concordance with timeline follow-back methods.³⁷ Participants who reported any smoking were then asked how many times they tried to quit smoking since the last assessment, the number of days smoked over the past 2 weeks, and when they last smoked. For Sample 1 (current smokers), primary dependent variables included the likelihood of (1) initiating a quit attempt, (2) achieving 7-day PPA (i.e., self-reported abstinence for the previous seven days at each assessment), and (3) reporting 7-day PPA at 2-month follow-up. For Sample 2 (recent quitters), primary dependent variables included likelihood of reporting a lapse (i.e., any instance of smoking following a quit attempt), relapse (i.e., seven consecutive days of smoking following a quit attempt), and 7-day PPA (at 2-month follow-up).

Data Analytic Plan

Analyses were conducted using SPSS Version 22. First, Cox regression survival analyses were employed to test past 2-week pain status as a predictor of initiating a quit attempt and achieving 7-day PPA among current smokers in Sample 1 and to lapse (i.e., first instance of smoking/puff) and relapse (i.e., seven consecutive days of smoking) among recent quitters in Sample 2. The Cox regression is a well-established statistical procedure that has frequently been used to examine predictors of cessation milestones, including smoking lapse/relapse outcomes.^25,38 This statistical test models the relationship between a set of covariates and the time to an event of interest, and provides an estimate of the hazard ratio (HR) and corresponding confidence interval (CI), which indicates the relative likelihood of an event occurring at any given point in time and accounts for both event status and time to event.³⁹ Consistent with previous work,^25,38 cases were censored if the individual did not report an “event” (i.e., did not initiate a quit attempt, achieve 7-day PPA, lapse, or relapse) or withdrew from the study before reporting an event. The Cox regression model uses a likelihood-based approach to estimation that utilizes all of the information available and adjusts for whether or not an individual is censored.⁴⁰ The number of cases censored for each outcome were as follows: n = 25 initiation of a quit attempt (n = 14 did not initiate a quit attempt [n = 11 with pain, n = 3 without pain], n = 11 lost to follow-up [n = 6 with pain, n = 5 without pain]), n = 131 achievement of 7-day PPA (n = 109 did not achieve 7-day PPA [n = 48 with pain, n = 61 without pain], n = 22 lost to follow-up [n = 9 with pain, n = 13 without pain]), n = 145 lapse (n = 113 did not lapse [n = 1 with pain, n = 112 without pain], n = 32 lost to follow-up [n = 0 with pain, n = 32 without pain]), and n = 163 relapse (n = 129 did not relapse [n = 5 with pain, n = 124 without pain], n = 34 lost to follow-up [n = 0 with pain, n = 34 without pain]). For ease of interpretation, the reference group was coded as the co-occurring pain group when examining associations between pain status and initiation of a quit attempt and achievement of 7-day PPA, whereas the reference group was coded as the no co-occurring pain group when examining associations between pain status and lapse and relapse.

Second, we employed logistic regression to examine 7-day PPA at the 2-month follow-up (for both Samples 1 and 2), with the reference group being comprised of those with co-occurring pain. For all primary outcomes, both unadjusted and adjusted models were tested. Adjusted models accounted for relevant sociodemographic and smoking factors (i.e., gender, race, relationship status, level of education, employment status, income, and cigarette dependence).^41–44 All logistic regression assumptions were met,⁴⁵ and the variance inflation factor for each predictor variable fell within the acceptable range (<10; Sample 1 range: 1.07–1.35, Sample 2 range: 1.07–1.35),⁴⁶ indicating that there were no issues related to multicollinearity. Finally, cross-sectional associations between pain status and confidence in quitting/remaining abstinent and nicotine withdrawal severity were examined using separate analysis of covariance models, and indirect associations between pain and latency to achieve cessation milestones via withdrawal/confidence were assessed using the PROCESS Macro for SPSS (controlling for the same sociodemographic and smoking factors).⁴⁷

Results

Participant Characteristics

Table 1 presents baseline characteristics for both samples as a function of past 2-week pain status. Sample 1 participants (N = 301 current smokers) were approximately 35 years old, smoked nearly one pack of cigarettes per day, and reported moderate levels of cigarette dependence. The majority of Sample 1 participants were white, male, in a relationship, employed at least part-time, and had completed more than a high school education. Approximately 23% of Sample 1 participants endorsed past 2-week pain. Current smokers with co-occurring pain (vs. no co-occurring pain) were more likely to be white and employed, reported smoking a greater number of cigarettes per day, and scored higher on a measure of cigarette dependence (ps < .05).

Table 1.

Sociodemographic, Smoking, and Pain Characteristics

	Sample 1			Sample 2
	No pain (n = 231)	Pain (n = 70)	Total sample (N = 301)	No pain (n = 166)	Pain (n = 76)	Total sample (N = 242)
	n (%)	n (%)	n (%)	n (%)	n (%)	n (%)
Racea,b
White	159 (68.8)	57 (81.4)	216 (71.8)	125 (75.3)	72 (94.7)	197 (81.4)
Non-White	72 (31.2)	13 (18.6)	85 (28.2)	41 (24.7)	4 (5.3)	45 (18.6)
Genderb
Male	158 (68.4)	49 (70.0)	207 (68.8)	112 (67.5)	73 (96.1)	185 (76.4)
Female	73 (31.6)	21 (30.0)	93 (31.2)	54 (32.5)	3 (3.9)	57 (23.6)
Relationship statusb
Single	78 (33.8)	19 (27.1)	97 (32.2)	83 (50.0)	2 (2.6)	85 (35.1)
In a relationship	153 (66.2)	51 (72.9)	204 (67.8)	83 (50.0)	74 (97.4)	157 (64.9)
Educationb
High school or less	58 (25.1)	10 (14.3)	68 (22.6)	20 (12.0)	1 (1.3)	21 (8.7)
Some college or more	173 (74.9)	60 (85.7)	233 (77.4)	146 (88.0)	75 (98.7)	221 (64.9)
Employmenta
Employed	206 (89.2)	54 (77.1)	260 (86.4)	159 (95.8)	74 (97.4)	233 (96.3)
Unemployed	25 (10.8)	16 (22.9)	41 (13.6)	7 (4.2)	2 (2.6)	9 (3.7)
Incomea,b
<$40 000 or unknown	121 (52.4)	23 (32.8)	144 (47.8)	64 (38.6)	2 (2.6)	66 (27.3)
≥$40 000	110 (47.6)	47 (67.1)	157 (52.2)	102 (61.4)	74 (97.4)	176 (72.7)
	M (SD)	M (SD)	M (SD)	M (SD)	M (SD)	M (SD)
Age	34.65 (8.56)	36.64 (11.93)	35.11 (9.47)	33.83 (8.44)	32.72 (4.71)	33.48 (7.48)
Cigarettes per daya,b	17.30 (4.87)	22.90 (6.66)	18.60 (5.83)	17.33 (6.34)	27.37 (5.23)	20.48 (7.60)
Cigarette dependencea	2.65 (1.06)	3.70 (0.82)	2.89 (1.10)	3.49 (1.03)	3.67 (0.74)	3.55 (0.95)

^aSignificant (p < .05) difference as a function of pain status among Sample 1 (current smokers).

^bSignificant (p < .05) difference as a function of pain status among Sample 2 (recent quitters).

Sample 2 participants (N = 242 recent quitters) were approximately 33 years old, smoked just over one pack of cigarettes per day, and reported moderate levels of nicotine dependence. Similar to Sample 1, participants in Sample 2 were mostly white, male, in a relationship, employed at least part-time, and had completed more than a high school education. Approximately 31% of Sample 2 participants endorsed past 2-week pain. Recent quitters with co-occurring pain (vs. no co-occurring pain) were more likely to be male, white, in a relationship, have completed at least some college, and make at least $40 000 per year (ps < .05). Recent quitters with co-occurring pain also reported smoking a greater number of cigarettes per day prior to quitting (p < .05).

Pain as a Predictor of Smoking Cessation Milestones

Initiation of a Quit Attempt

Over 90% (n = 276) of current smokers (Sample 1) initiated a quit attempt during the study period. Nearly all smokers without co-occurring pain (n = 223, 96.5%; Figure 1) initiated a quit attempt during the study period (median time to initiation = 2 weeks), compared with only three quarters (n = 53, 75.7%) of those with co-occurring pain (median time to initiation = 4 weeks). Examination of the hazard ratio indicated that current smokers who reported no past 2-week pain were over two times as likely to initiate a quit attempt at any time during the study period, relative to those who endorsed past 2-week pain (HR = 2.22, CI: 1.60–3.08, p < .001; Table 2). Past 2-week pain status remained a significant predictor of initiating a quit attempt, even after accounting for covariates (adjusted HR = 1.94, CI: 1.35–2.77, p < .001; Table 2).

Figure 1.

Percentage of participants who achieved cessation milestones as a function of past 2-wk pain status. Initiation of quit attempt, achievement of 7-d PPA, and 7-d PPA at follow-up were assessed among Sample 1 (a), and lapse, relapse, and 7-d PPA at follow-up were assessed among Sample 2 (b); *p < .01.

Table 2.

Sample 1: Associations Between Pain Status and Cessation Milestones

	Unadjusted hazard ratio (95% confidence interval)	p	Adjusted hazard ratio (95% confidence interval)	p
Quit attempt
Gender
Male	1.121 (0.867–1.451)	.383	0.936 (0.707–1.239)	.645
Race
Non-white	1.212 (0.931–1.579)	.153	1.135 (0.853–1.511)	.385
Relationship status
In a relationship	1.076 (0.835–1.388)	.570	0.972 (0.737–1.281)	.840
Education status
High school or less	1.202 (0.906–1.595)	.203	1.115 (0.826–1.505)	.478
Employment status
Nonemployed	0.912 (0.633–1.312)	.619	0.865 (0.567–1.321)	.503
Income
≥$40k	0.830 (0.654–1.055)	.129	1.148 (0.881–1.496)	.307
Heaviness of smoking	0.847 (0.756–0.948)	.004**	0.907 (0.804–1.024)	.116
Pain status
No pain	2.219 (1.602–3.075)	<.001**	1.937 (1.354–2.772)	<.001**
Achievement of abstinencea
Gender
Male	0.680 (0.493–0.937)	.018*	0.800 (0.563––1.135)	.211
Race
Non-white	2.115 (1.544–2.898)	<.001**	1.757 (1.254–2.460)	.001**
Relationship status
In a relationship	1.607 (1.169–2.208)	.003**	1.332 (.934–1.900)	.113
Education status
High school or less	2.473 (1.761–3.473)	<.001**	2.009 (1.392–2.901)	<.001**
Employment status
Nonemployed	0.724 (0.432–1.212)	.219	0.585 (0.327–1.048)	.657
Income
≥$40k	1.211 (0.894–1.639)	.216	0.770 (0.548–1.082)	.132
Heaviness of smoking	0.614 (0.526–0.718)	<.001**	0.697 (0.595–0.816)	<.001**
Pain status
No pain	4.557 (2.584–8.035)	<.001**	3.178 (1.739–5.806)	<.001**
7-d PPA at follow-upb
Gender
Male	0.979 (0.601–1.595)	.931	0.967 (0.496–1.882)	.920
Race
Non-white	0.506 (0.301–0.851)	.010**	2.113 (1.067–4.186)	.032*
Relationship status
In a relationship	1.048 (0.645–1.701)	.851	0.882 (0.461–1.686)	.704
Education status
High school or less	2.638 (1.475–4.719)	.001**	2.649 (1.237–5.671)	.012*
Employment status
Nonemployed	2.152 (1.089–4.251)	.027*	0.628 (0.252–1.563)	.317
Income
≥$40k	1.384 (0.878–2.179)	.161	0.665 (0.370–1.196)	.173
Heaviness of smoking	0.397 (0.304–0.518)	<.001**	0.471 (0.348–0.638)	<.001**
Pain status
No pain	9.564 (4.761–19.212)	<.001**	5.194 (2.361–11.426)	<.001**

^a7-d point prevalence abstinence.

^b7-d point prevalence abstinence at the 2-month follow-up assessment.

*p < .05; **p ≤ .01.

Achievement of 7-Day PPA

Approximately 57% (n = 170) of current smokers (Sample 1) achieved 7-day PPA during the study period. Over two thirds (n = 157, 68%; Figure 1) of smokers without co-occurring pain achieved abstinence over the course of the study (median time to 7-day PPA = 6 weeks), compared with only 18.6% (n = 13) of those with co-occurring pain (median time to 7-day PPA = 8 weeks). Examination of the hazard ratio indicated that current smokers who reported no past 2-week pain were 4.5 times as likely to achieve 7-day PPA, relative to those who endorsed past 2-week pain (HR = 4.56, CI: 2.58–8.04, p < .001; Table 2). Past 2-week pain status remained a significant predictor of 7-day PPA, even after accounting for covariates (adjusted HR = 3.18, CI: 1.74–5.81, p < .001; Table 2).

Lapse

Approximately 40% (n = 97) of recent quitters (Sample 2) endorsed a smoking lapse during the study period. Nearly all (n = 75, 98.7%; Figure 1) recent quitters with co-occurring pain reported a lapse to smoking (median time to lapse = 2 weeks), compared with just 13.3% (n = 22) of those without co-occurring pain (median time to lapse = 8 weeks). Indeed, past 2-week pain status was a significant predictor of smoking lapse (HR = 11.09, CI: 6.44–19.09, p < .001; Table 3). Even after accounting for covariates, the hazard ratio indicated that recent quitters who endorsed past 2-week pain were nearly 10 times as likely to report a lapse, relative to those without pain (adjusted HR = 9.71, CI: 5.02–18.77, p < .001; Table 3).

Table 3.

Sample 2: Associations Between Pain Status and Cessation Milestones

	Unadjusted hazard ratio (95% confidence interval)	p	Adjusted hazard ratio (95% confidence interval)	p
Lapse
Gender
Female	0.488 (0.272–0.876)	.016*	0.759 (0.326–1.765)	.522
Race
Non-white	0.434 (0.218–0.862)	.017*	0.865 (0.344–2.175)	.759
Relationship status
In a relationship	0.245 (0.134–0.449)	<.001**	0.792 (0.361–1.741)	.562
Education status
High school or less	0.425 (0.156–1.157)	.094	1.033 (0.326–3.274)	.956
Employment status
Nonemployed	1.570 (0.638–3.862)	.326	1.732 (0.544–5.510)	.352
Income
≥$40k	0.254 (0.128–0.506)	<.001**	0.738 (0.310–1.754)	.491
Heaviness of smoking	1.131 (0.909–1.409)	.270	1.045 (0.808–1.352)	.737
Pain status
Pain	11.086 (6.439–19.085)	<.001**	9.707 (5.021–18.766)	<.001**
Relapse
Gender
Female	0.163 (0.060–0.446)	<.001**	1.239 (0.255–6.027)	.791
Race
Non-white	0.164 (0.052–0.521)	0.002**	0.302 (0.058–1.583)	.157
Relationship status
In a relationship	0.067 (0.021–0.213)	<.001**	0.304 (0.081–1.148)	.079
Education status
High school or less	0.125 (0.017–0.898)	.039*	0.637 (0.066–6.142)	.697
Employment status
Nonemployed	0.719 (0.177–2.926)	.645	0.822 (0.096–7.004)	.858
Income
≥$40k	0.097 (0.031–0.308)	<.001**	0.997 (0.220–4.514)	.996
Heaviness of smoking	1.226 (0.957–1.571)	.107	1.090 (0.792–1.501)	.597
Pain status
Pain	33.245 (15.256–72.445)	<.001**	27.102 (10.288–71.392)	<.001**
7-d PPA at follow-upb
Gender
Female	3.517 (1.823–6.785)	<.001**	0.547 (0.228–1.314)	.177
Race
Non-white	2.476 (1.242–4.938)	.010*	0.991 (0.388–2.527)	.984
Relationship status
In a relationship	2.165 (1.259–3.724)	.005**	0.556 (0. 0.248–1.247)	.154
Education status
High school or less	3.348 (1.186–9.456)	.023*	1.468 (0.393–5.479)	.568
Employment status
Nonemployed	0.463 (0.113–1.895)	.284	0.277 (0.054–1.418)	.123
Income
≥$40k	1.546 (0.872–2.742)	.136	0.585 (0.261–1.310)	.192
Heaviness of smoking	0.737 (0.559–0.972)	.031*	0.815 (0.556–1.193)	.293
Pain status
No pain	46.957 (16.225–135.895)	<.001**	66.787 (20.955–212.861)	<.001**

^a7-d point prevalence abstinence.

^b7-d point prevalence abstinence at the 2-month follow-up assessment.

*p < .05; **p < .01.

Relapse

Approximately 33% (n = 79) of recent quitters (Sample 2) endorsed a relapse to smoking during the study period. Again, nearly all (n = 71, 93%; Figure 1) recent quitters with co-occurring pain reported a relapse to smoking (median time to relapse = 2 weeks), compared with just 4.8% (n = 8) of those without co-occurring pain (median time to relapse = 8 weeks). Past 2-week pain status was a significant predictor of relapse to smoking (HR = 33.25, CI: 15.26–72.45, p < .001; Table 3). Even after accounting for covariates, the hazard ratio indicated that recent quitters who endorsed past 2-week pain remained over 20 times as likely to report a relapse (adjusted HR = 27.10, CI: 10.29–71.39, p < .001; Table 3).

Seven-Day PPA at Follow-up

Among Sample 1, approximately 53% (n = 159) of participants reported 7-day PPA at the 2-month follow-up. The absence of pain was associated with a 10 times greater odds of reporting 7-day PPA at follow-up (odds ratio [OR] = 9.56, CI: 4.76–19.21, p < .001; Table 2). As shown in Figure 1, only 15.7% of participants (n = 11) who endorsed past 2-week pain were abstinent at follow-up, compared with 64.1% of those who did not endorse past 2-week pain (n = 148). Even after accounting for covariates, current smokers without co-occurring pain had a five times greater odds of achieving 7-day PPA (adjusted OR = 5.19, CI: 2.36–11.43, p < .001). However, the Hosmer and Lemeshow chi-square indicated poor model fit (Hosmer and Lemeshow χ ² = 18.93, p = .02), precluding interpretation of this result.

Among Sample 2, approximately 51% (n = 124) of participants reported 7-day PPA at the 2-month follow-up. The odds that recent quitters without co-occurring pain reported 7-day PPA at follow-up were approximately 47 times greater, relative to those with co-occurring pain (OR = 46.96, CI: 16.23–135.90, p < .001; Table 2). As shown in Figure 1, only 5% (n = 4) of participants who endorsed past 2-week pain were abstinent at follow-up, compared with 72.3% of those who did not endorse past 2-week pain (n = 120). After accounting for covariates, recent quitters without co-occurring pain had a 66 times greater odds of achieving 7-day PPA (adjusted OR = 66.79, CI: 20.96–212.86, p < .001; Hosmer and Lemeshow χ ² = 7.85, p = .35).

Potential Mechanisms

Confidence in Quitting/Remaining Quit

Participants with pain (vs. without pain) reported lower confidence in successfully quitting smoking. This was observed among both Sample 1 (M = 5.65, SE = 0.31 vs. M = 7.69, SE = 0.20; F(1, 292) = 38.585, η _p² = 0.12, p < .001) and Sample 2 (M = 2.97, SE = 0.38 vs. M = 7.93, SE = 0.32; F(1, 233) = 341.75, η _p² = 0.60, p < .001). Among Sample 1, there were significant indirect associations between pain status and latency to both initiating a quit attempt (b = 0.518 [SE = 0.152], 95% CI [0.272 to 0.882]) and achieving 7-day PPA (b = 0.650 [SE = 0.225], 95% CI [0.286 to 1.182]) via confidence in quitting smoking. Specifically, positive pain status was associated with lower reported confidence in quitting, which, in turn, was associated with a longer duration to initiating a quit attempt and achieving abstinence. Similarly, among Sample 2, there were significant indirect associations between pain status and latency to both lapse (b = −1.674 [SE = 0.536], 95% CI [−2.677 to −0.542]) and relapse (b = −1.856 [SE = 0.532], 95% CI [−2.780 to −0.734]) via confidence in remaining quit. Specifically, positive pain status was associated with lower reported confidence in remaining quit at the baseline assessment, which, in turn, was associated with a shorter duration to first lapse and relapse.

Nicotine Withdrawal Severity

Baseline levels of nicotine withdrawal severity were only examined among recent quitters (Sample 2), given that current smokers (Sample 1) had not yet initiated a quit attempt. Recent quitters (Sample 2) with pain reported more severe nicotine withdrawal at the baseline assessment compared with those without pain (M = 2.85, SE = 0.16 vs. M = 1.24, SE = 0.13; F(1, 233) = 214.91, η _p² = .48, p < .001). We also observed an indirect association between pain status and latency to lapse via nicotine withdrawal severity (b = −0.861 [SE = 0.394], 95% CI [−1.592 to −0.034]), such that pain was associated with greater withdrawal at baseline, which, in turn, was associated with a shorter duration to lapse. The indirect association between pain status and latency to relapse via withdrawal severity was not statistically significant (p > .05).

Discussion

To our knowledge, this is the first study to demonstrate that the presence of co-occurring pain prospectively predicts poor outcomes across a range of smoking cessation milestones. Specifically, participants with past 2-week pain (vs. no pain) were less likely to initiate a quit attempt and achieve 7-day PPA and were more likely to endorse a lapse and/or relapse to smoking during the 2-month study period. Sample 2 participants with past 2-week pain were also less likely to report 7-day PPA at the 2-month follow-up. Importantly, these effects were observed after accounting for several relevant covariates (i.e., gender, race, relationship status, level of education, employment status, income, and cigarette dependence), and some effect sizes were surprisingly large in magnitude (adjusted hazard/odds ratios ranged from approximately 2.0–67.0). Across both samples, participants with pain also reported more severe nicotine withdrawal and lower confidence in quitting/staying quit at baseline, and analysis of indirect associations suggests that these factors may play a mechanistic role in the effects of pain status on smoking cessation outcomes.

These results contribute to a growing empirical literature indicating that pain contributes to the maintenance of addiction in general¹ and tobacco cigarette smoking in particular.⁴⁸ Previous research has found that pain-related factors can motivate smoking, influence pre-cessation processes, and predict lapse/relapse behavior.^{2,10,14,15,22–25} The current study expands on this work by providing novel prospective evidence that smokers with co-occurring pain (vs. without co-occurring pain) are less likely to initiate a quit attempt and maintain smoking abstinence. Collectively, the current findings suggest that smokers with pain likely face unique barriers to successfully quitting smoking and highlight the potential utility of assessing and addressing pain among all tobacco cigarette smokers.

One clinical implication of these findings is that smoking cessation interventions may benefit from tailoring to account for the antithetical influence of pain.^49–51 Cessation interventions have been successfully administered to smokers with chronic pain,^52–54 and treatments that incorporate pain-smoking psychoeducation and provide smokers with alternative pain-coping skills training could potentially improve outcomes. For example, a brief integrated treatment that was developed to address smoking in the context of chronic pain⁵³ showed that smokers who received the tailored intervention (vs. control) were 2.5 times more likely to report an interest in learning about cessation programs and five times more likely to endorse willingness to consider an intensive smoking cessation program. Previous cross-sectional research has further shown that smokers with pain (vs. no pain) tend to report lower cessation self-efficacy¹⁵ and anticipate more severe nicotine withdrawal during future quit attempts.¹⁶ Thus, future treatments should aim to increase self-efficacy for quitting and remaining abstinent in the context of pain.⁵⁵ Smokers with pain may also benefit from high-dose or combination nicotine replacement therapy.^22,56,57

Several limitations should be noted. First, data were collected using online survey methods and smoking status/abstinence could not be verified. Future research should include biochemical verification of smoking status and abstinence (e.g., via exhaled carbon monoxide).⁵⁸ Second, past 2-week pain was assessed using a single item. Although this item has been positively associated with established pain measures,³¹ future prospective work would benefit from incorporating more comprehensive assessment of pain. Third, we observed wide confidence intervals in several of the models (e.g., Sample 2 logistic regression model), which may indicate a lack of precision in the sample estimates, and should serve as a warning against overinterpreting the results.⁵⁹ Several factors may lead to wide confidence intervals, including small sample/cell sizes and multicollinearity problems.^59,60 Although the variance inflation factor for each predictor indicated a low probability of multicollinearity, we did observe some issues related to sample size (e.g., only n = 4 recent quitters with pain reported 7-day PPA at follow-up), and future research is needed to replicate these findings among larger samples. Fourth, both samples were predominantly male, white, well-educated, partnered, and employed, and future iterations should include more diverse samples in terms of gender, race/ethnicity, marital status, and education/employment status, and should test whether sociodemographic factors moderate prospective associations between pain and cessation milestones. For example, recruitment of a more gender-balanced sample would permit examination of potential sex differences. Fifth, participants were not provided with pharmacological or behavioral cessation aids. Although nearly 70% of smokers who attempt to quit do not utilize recommended cessation aids (e.g., nicotine replacement, cognitive-behavioral counseling),⁶¹ the extent to which these results may generalize to smokers receiving cessation treatment remains unclear. Finally, the use of cross-sectional data at baseline prohibits causal inferences regarding the role of confidence in quitting and nicotine withdrawal in the effects of pain status on cessation outcomes. Research has shown that cessation self-efficacy dynamically changes over the course of quit attempt,^18,62 and future research would benefit from using prospective methods (e.g., ecological momentary assessment⁴⁸) to permit tests of temporal ordering and further clarify potential mechanisms of action.

In summary, this study demonstrated that the presence of co-occurring pain prospectively predicts poorer outcomes across a range of cessation milestones. Limited research has examined the effects of co-occurring pain on smoking cessation, and these findings represent an important step toward better understanding the role of pain in the maintenance of cigarette dependence. This and future work has the potential to inform the development of tailored interventions for smokers for co-occurring pain.

Supplementary Material

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

Funding

This work was supported by the National Institute on Drug Abuse (NIDA) of the National Institutes of Health (K12 DA031794 and K23 DA041616) and by the Chairman’s Research Development Fund Pilot Grant Program, Department of Psychiatry and Behavioral Sciences, Medical University of South Carolina.

Declaration of Interests

None declared.