- Email: [email protected]
Cigarette smoking and pain: Depressive symptoms mediate smoking-related pain symptoms
Ò
PAIN 153 (2012) 1749–1754
www.elsevier.com/locate/pain
Cigarette smoking and pain: Depressive symptoms mediate smoking-related pain symptoms Jenna Goesling ⇑, Chad M. Brummett, Afton L. Hassett Department of Anesthesiology, University of Michigan, Ann Arbor, MI, USA
Sponsorships or competing interests that may be relevant to content are disclosed at the end of this article.
a r t i c l e
i n f o
Article history: Received 5 January 2012 Received in revised form 20 April 2012 Accepted 11 May 2012
Keywords: Smoking Chronic pain Depression Smoking cessation
a b s t r a c t Numerous studies have shown an association between smoking and pain, with smokers reporting more pain and worse functioning. However, little is known about factors that impact this complex relationship. This study investigated the association between smoking, pain, and depressive symptoms. Participants were new patients seen at a multidisciplinary pain clinic. All patients were mailed an intake packet of validated questionnaires as part of an ongoing research and clinical care initiative. Of the 497 patients evaluated, 426 had valid smoking data. Among these patients, 32.6% (n = 139) reported being current smokers, 31.7% (n = 135) were classified as former smokers, and 35.7% (n = 152) were never smokers. A multivariate analysis of covariance (smoking status, age, gender, education) revealed a main effect for pain severity (F = 7.36, P < 0.001), pain interference (F = 4.03, P = 0.001), and depressive symptoms (F = 7.87, P < 0.001). Current smokers demonstrated higher pain severity, pain interference, and depressive symptoms compared with former smokers and never smokers (P < 0.01 for all analyses), while there were no differences between the former-smoker and never-smoker groups. However, the effect of smoking on pain severity (P = 0.06) and pain interference (P = 0.22) was no longer significant after controlling for depressive symptoms in a mediation model. Additionally, among former smokers, longer quit duration was associated with less pain severity. In conclusion, smoking rates were high and smoking was associated with a worse chronic pain phenotype. Importantly, depressive symptoms emerged as a critical mediating factor in helping to explain the relationship between smoking and pain. Ó 2012 International Association for the Study of Pain. Published by Elsevier B.V. All rights reserved.
1. Introduction Smoking and chronic pain are recognized as 2 of the most challenging public health concerns. It is estimated that 20.6% of U.S. adults are current cigarette smokers [7]. The majority of studies with pain populations find that smoking rates are higher than the population rate [14,25]. For example, 54% of chronic low back pain patients reported being current smokers [25]. In addition to establishing smoking rates among pain patients, researchers have begun to characterize differences between smokers and nonsmokers with regards to pain-related complaints [23,36,37]. The majority of studies show that smoking is related to more pain [36,37]. In a population-based study, current smokers and former heavy smokers reported more pain locations and greater pain intensity compared to people who never smoked [27]. Similarly, data from
⇑ Corresponding author. Address: Back & Pain Center, Burlington Building 1, Suite 100, 325 E. Eisenhower Parkway, Ann Arbor, MI 48108, USA. Tel.: +1 734 232 1265; fax: +1 734 936 6585. E-mail address: [email protected] (J. Goesling).
the National Spine Institute found that in those with chronic back pain, smokers reported more severe pain than nonsmokers [36]. Several hypotheses have been advanced in attempting to unravel the nature of the association between smoking and pain. One possibility is that the aversive physiological effects of smoking cause or aggravate painful conditions [35]. However, it has been challenging to test this hypothesis because existing data are generally cross-sectional [30], and because there are many precipitating factors for pain. Thus, it is difficult to isolate causal relationships unless they are particularly strong. A meta-analysis of 41 studies on smoking and low back pain found that even when a positive association was found, effect sizes were small [30], suggesting that the evidence that smoking directly causes low back pain is weak. An alternative hypothesis is that smokers and people who have chronic pain share biological or environmental features. One variable of interest is depression, given its robust association with smoking [11,33] and pain [2,13]. Only one study has considered the multivariate associations between smoking, depression, and pain [21]. This study showed that smokers reported more pain but this association weakened when controlling for depression, suggesting that depression may in part explain why smoking is
0304-3959/$36.00 Ó 2012 International Association for the Study of Pain. Published by Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.pain.2012.05.014
1750
Ò
J. Goesling et al. / PAIN 153 (2012) 1749–1754
related to pain. Given the relatively weak evidence for a direct causal relationship between smoking and pain, further insight into comorbidities such as depression would be quite valuable. The present study was intended to address this issue in depth. The first aim of the current study was 2-fold: to establish smoking rates in a tertiary pain clinic and investigate the effects of smoking status on measures of pain and depressive symptoms. Specifically, it was hypothesized that smoking rates in our patient sample would be higher than the population rate, and smokers would report greater pain severity and pain interference as well as more symptoms of depression. The second aim was to test a mediation model specifically considering the role of depressive symptoms. It was predicted that depressive symptoms would partially account for the effect of smoking on pain measures. Finally, the third aim was to explore differences between former and current smokers. It was hypothesized that former smokers would have less pain and fewer symptoms of depression than current smokers. 2. Methods 2.1. Study setting and participants The Back & Pain Center (Department of Anesthesiology, University of Michigan Health System) is a multi-disciplinary outpatient pain medicine clinic where physicians treat a range of pain conditions and evaluate approximately 1400 new patients per year. Prior to the initial visit, all new patients are mailed a battery of validated questionnaires as part of an ongoing research and clinical care initiative to phenotype patients. A cover letter is included stating that some of the information from the questionnaires may be used for research, and a waiver of informed consent is obtained. It is stated in the cover letter that no identifying data will be disclosed should their information be used for research purposes. Patient packets are collected at the initial visit and entered into the Assessment of Pain Outcomes Longitudinal Electronic Data Capture system and exported for data analysis [17]. The collection of data and all subsequent analyses are approved by the Institutional Review Board at the University of Michigan (Ann Arbor, MI, USA). All new patients evaluated at the clinic from November 2010 through May 2011 who completed a new patient questionnaire were included in the initial analysis. Of the 497 patients evaluated to date, 71 could not be classified into 1 of the 3 smoking groups due to missing or incomplete data on the smoking items; therefore, they were excluded from the analyses. There were no other exclusion criteria. This resulted in 426 patients with complete smoking data who were included in the study. 2.2. Measures 2.2.1. Pain severity and pain interference The Brief Pain Inventory is a widely used, validated self-report scale that assesses the severity of pain and impact of pain of daily functioning [9]. Pain severity is assessed using a 4-item subscale that asks about pain at its worst, least, and average in the last week, and pain right now. For each item, patients rate their pain on a scale of 0-10 (0 = ‘‘No pain’’ and 10 = ‘‘Pain as bad as you can imagine’’). The average of the 4 scores can be used as a single composite measure of pain severity [26]. Pain interference is assessed using a 7-item subscale. Patients are asked to rate how much their pain interfered with different areas of their life (general activity, mood, walking ability, normal work, relations with others, sleep, and enjoyment of life) in the past week on a scale of 0-10 (0 = ‘‘Does not interfere’’ and 10 = ‘‘Completely interferes’’).
2.2.2. Depressive symptoms The Hospital Anxiety and Depression Scale is a brief and widely used instrument to measure psychological distress in general and medical populations [38]. It includes 2 7-item subscales that assess for anxiety and depressive symptoms. For the purpose of this study, we included only the depression subscale. The possible scores on this subscale range from 0 to 21. A score of 0-7 is considered within the normal range; a score of 8-10 is suggestive of the presence of depression; and a score of 11 or higher indicates that there is a high probability that depression is present [38]. 2.2.3. Smoking variables Patients were classified into 3 groups: (1) current smoker, (2) former smoker, and (3) never smoker. Smoking status was determined based on the question ‘‘Do you smoke?’’ (Yes = current smoker, No = current nonsmoker). Smoking history was assessed using the question, ‘‘Have you ever been a smoker?’’ Current nonsmokers who said, ‘‘yes’’ to this question were classified as former smokers. Former smokers provided additional information about how long it had been since they quit smoking, and number of years quit was calculated. If a patient said ‘‘no’’ to both questions (‘‘Do you smoke?’’ and ‘‘Have you ever been a smoker?’’), he/she was classified as a never smoker. Additionally, amount smoked was measured using current smokers’ response to the question, ‘‘How many packs per day?’’ 2.3. Data analysis Data were analyzed using SPSS (version 18; SPSS Inc, Chicago, IL, USA). Demographic characteristics (age, gender, education) were summarized in order to evaluate relevant characteristics of the sample. Mean and SD were reported for the continuous variable (age) and proportions were reported for categorical variables (gender and education), using analysis of variance and v2, respectively. Multivariate analysis of covariance was utilized to determine if smoking status was associated with pain severity, pain interference, and depressive symptoms. Due to differences in age, gender, and education across smoking groups, all 3 demographic variables were used as covariates in the above analyses. To test the hypothesis that depressive symptoms would mediate the effect of smoking on both pain severity and pain interference, we applied 2 convergent mediation procedures. The first, widely used procedure requires that 4 steps be satisfied to determine mediation [3]. First, a regression analysis was used to demonstrate an association between smoking status (current smoker: yes/no) and the outcome variables (pain). Second, another regression analysis was run to show an association between smoking status and the proposed mediator (depressive symptoms). Third, to assess the association between the proposed mediator (depressive symptoms) and the outcome variable (pain), a regression model was run. Lastly, pain was simultaneously regressed on both smoking status and depressive symptoms to determine if the effect of smoking on pain was reduced after adjusting for depressive symptoms. These 4 steps were done separately for pain severity and pain interference. An additional analysis was performed to address a potential limitation in the mediation results (one of the items on the Brief Pain Inventory pain interference scale measures mood, which is likely to overlap with depressive symptoms and could confound the model). We created a 6-item pain interference scale that excluded the mood item and reran the steps of the mediation analysis using this revised scale. The results did not change after the mood item was removed, so the model with the original validated 7-item scale is reported. For the second mediation procedure, we used a single, inclusive test of mediation (Sobel) as recommended by MacKinnon and colleagues [31]. The Sobel test
Ò
1751
J. Goesling et al. / PAIN 153 (2012) 1749–1754
[F(2, 423) = 20.94, P < 0.001]. Post hoc tests were performed and revealed that former smokers were more likely to be older than both current smokers and never smokers, Ps < 0.001. Current and never smokers did not differ on age, P = 0.24. Finally, the overall v2 was significant for education: v2 (2) = 17.57, P < 0.001. Current smokers differed from both former and never smokers (Ps < 0.0001) in that they were less likely to have graduated from college. Former and never smokers did not differ in regard to education (P = 0.11).
Table 1 Demographic differences between smokers, former smokers, and never smokers. Current smokers (n = 139)
Sex (male) Age College graduate (yes)
46.8%a 43.42a (12.02) 20.9%a
Nonsmokers Former smokers (n = 135)
Never smoked (n = 152)
51.5%a 55.44b (17.14) 35.1%b
32.9%b 46.7a (17.72) 44.5%b
Overall P-value
0.004 <0.0001 <0.0001
3.3. Effect of smoking on pain severity, pain interference, and depressive symptoms
Data presented as percentages or mean (SD). Values that do not share subscripts are significantly different at P < 0.05.
provides a conservative test of the indirect effect of the independent variable on the dependent variable via the mediator. It is often used as an additional test of mediation to complement the 4-step procedure described above. A significant Sobel test would suggest that the effect of smoking status on pain is largely driven by the intervening variable of depressive symptoms [31]. 3. Results 3.1. Smoking characteristics Among the 426 patients with valid smoking data, 32.6% (n = 139) reported being current smokers, 31.7% (n = 135) were classified as former smokers, and 35.7% (n = 152) were never smokers; 47.7% of smokers reported smoking a pack or more a day. Amount smoked per day was regressed on all 3 outcome variables in separate regression analyses. There was no effect for amount smoked on any of the outcome variables: pain severity (b = 0.06, t = 0.250, P = 0.80), pain interference (b = 0.05, t = 0.167, P = 0.868), or depressive symptoms (b = 0.62, t = 0.985, P = 0.326). The majority of former smokers (70.5%) reported having quit smoking over 5 years ago. Among the former smokers, number of years quit was regressed on all 3 outcome variables in separate regression analyses. There was no effect for number of years quit on pain interference (b = 0.005, t = 0.345, P = 0.730) or depressive symptoms (b = 0.021, t = 0.719, P = 0.473). However, there was an effect of number of years quit on pain severity (b = 0.028, t = 2.24, P = 0.027), with a shorter duration of being quit associated with higher pain severity. 3.2. Demographic characteristics The demographic characteristics of the participants are provided in Table 1. The overall v2 test for gender was significant: v2 (2) = 11.05, P = 0.004. Post hoc tests revealed there were no differences between current and former smokers with regard to gender (P = 0.43); however, patients who had never smoked were more likely to be female compared to both smoking groups (P < 0.001). There was also a significant main effect for age
Table 2 shows the effect of smoking on pain severity, pain interference, and depressive symptoms. A multivariate analysis of covariance with smoking status predicting all 3 outcome variables revealed a main effect for pain severity [F(5, 383) = 7.36, P < 0.000], pain interference [F(5, 383) = 4.03, P = 0.001], and depressive symptoms [F(5, 383) = 7.87, P < 0.001], with current smokers reporting greater pain intensity, more pain interference, and more symptoms of depression. Pairwise comparisons (Fisher’s least significant difference) revealed that these differences were significant for current smokers compared to former smokers (pain severity P = 0.008; pain interference P = 0.004; depressive symptoms P < 0.001) and never smokers (pain intensity P = 0.001; pain interference P = 0.001; depressive symptoms P < 0.001), but not between former smokers and never smokers (pain intensity P = 0.56; pain interference P = 0.65; depressive symptoms P = 0.78). 3.4. Mediation model First, to test the association between the primary predictor (smoking status: yes/no) and the first outcome variable (pain severity), pain severity was regressed on smoking status; b = 0.374, t = 3.52, P < 0.001. Second, to determine if there was an effect for smoking status on the proposed mediator, depressive symptoms was regressed on smoking status; b = 1.50, t = 5.58, P < 0.001. Next, the effect of the mediator, depressive symptoms, on pain severity was tested; b = 0.15, t = 8.53, P < 0.001. Finally, to test whether depressive symptoms mediated the relationship between smoking and pain, pain severity was regressed on smoking status, controlling for depressive symptoms; b = 0.197, t = 1.87, P = 0.06. The reduction in significance of smoking status when controlling for depressive symptoms showed that depressive symptoms mediated the effect of smoking on pain severity (Fig. 1). To further support this conclusion, we tested the indirect effect using a single (Sobel) test [31] and again found evidence for mediation, Sobel (Aroian) test = 4.44, P < 0.001. We tested mediation for the second outcome measure, pain interference, using the same approach. To test the association between the primary predictor (smoking status) and the second outcome variable (pain interference), pain interference was regressed on smoking status; b = 0.58, t = 4.32, P < 0.001. The second step, the effect of smoking status on the proposed mediator, was identical to
Table 2 Current smokers report more pain severity, pain interference, and depressive symptoms compared to former and never smokers. Current smokers
BPI pain severity BPI pain interference HADS depression subscale
6.90 (1.64) 7.51 (2.06) 11.20 (4.35)
Nonsmokers Former smokers
Never smoked
6.30 (1.88) 6.62 (2.24) 8.50 (4.45)
6.17 (1.87) 6.43 (2.48) 8.21 (4.60)
Current vs former P value
Current vs never P value
Former vs never P value
0.008 0.004 <0.001
0.001 0.001 <0.001
0.56 0.65 0.78
Data presented as mean (SD). BPI, Brief Pain Inventory: Pain severity (0-10) 0 = No pain, 10 = Pain as bad as you can imagine; Pain interference: (0-10) 0 = does not interfere, 10 = completely interferes; HADS, Hospital Anxiety and Depression scale (0-21); higher values indicate more depression.
Ò
1752
J. Goesling et al. / PAIN 153 (2012) 1749–1754
Depression
.15*
1.50*
Smoking
Pain Severity .20 (.37*)
Fig. 1. Path diagram demonstrating that depressive symptoms mediated the effect of smoking on pain severity. Reported statistics are unstandardized regression coefficients. ⁄P < 0.05.
Depression
.31*
1.50*
Smoking
.14 (.58*)
Pain Interference
Fig. 2. Path diagram demonstrating that depressive symptoms mediated the effect of smoking on pain interference. Reported statistics are unstandardized regression coefficients. ⁄P < 0.05.
that shown in the previous mediation analysis; b = 1.50, t = 5.58, P < 0.001. Next, the effect of depressive symptoms on pain interference was tested; b = 0.31, t = 16.63, P < 0.001. Finally, to test whether depressive symptoms mediated the relationship between smoking and pain, pain interference was regressed on smoking status, controlling for depressive symptoms; b = 0.139, t = 1.22, P = 0.22. The reduction in significance when controlling for depressive symptoms shows that depressive symptoms mediated the effect of smoking on pain interference (Fig. 2). Again, we tested the indirect effect using a single (Sobel) test and found evidence of mediation, Sobel (Aroian) test = 5.22, P < 0.001.
sion in smokers is an important precursor to the identification of the particular mechanisms that lead patients who smoke to report worse pain [10,28]. For example, cognitive models of depression propose that, because smokers are more likely to be depressed, smokers may make more negative appraisals as a result of believing they have less control over their pain [10,28]. Another possibility is that, because depression and pain are thought to share complex neurophysiological pathways [6,12,13,34], depressed smokers may experience more pain due to this interaction. Our mediation findings are consistent with Hooten and colleagues, who reported that the association between smoking and pain severity was reduced when depression was included in the statistical model [21]. In addition to testing such a model in a different treatment setting and on a more heterogeneous population of pain patients, the current study showed a relatively stronger mediating effect for depressive symptoms on pain severity and complete mediation on a pain outcome (pain interference) not previously tested. The consistency of results across these 2 studies highlights the generalizability of the mediation model for these variables. It follows that when exploring the association between smoking and pain, researchers should consider the role of depression in their participant samples. Interestingly, although many prior studies have found an association between smoking and pain [27,36], others have not [14]. One possible explanation for these conflicting findings may be patient-level differences on depression among various pain populations. For example, smokers with pain in the general population may be different from smokers being treated at pain clinics. In support of this possibility, research suggests that pain clinic patients tend to be more depressed than those in the community [2]. While we cannot make definitive causal statements based on these data, our results have tangible clinical value. The presence of depressive symptoms among current smokers being treated at the clinic was particularly striking, suggesting that smokers being treated for pain should also receive an evaluation for depression. Smokers with chronic pain exhibiting depression would likely benefit from cognitive-behavioral therapy and/or antidepressants in addition to current pain treatment [5,16] to help manage their depressive symptoms and because cognitive-behavioral therapy is also indicated for the treatment of chronic pain [32], and some classes of antidepressants have analgesic effects [1,15,18]. 4.2. Smoking rates and phenotypic characteristics of smokers
4. Discussion Given the impact of both cigarette smoking and chronic pain on the health and well-being of millions of Americans, understanding their relationship is both scientifically and clinically important. Our data suggest that smoking rates in patients being treated at an outpatient clinic were high, and smoking was associated with a worse chronic pain phenotype. Importantly, symptoms of depression emerged as a critical factor in helping to give a clearer interpretation of the relationship between smoking and pain. 4.1. Depressive symptoms mediated the relationship between smoking and pain A key finding from this study was that depressive symptoms mediated the relationships between smoking and pain severity as well as pain interference, suggesting that the link between smoking and pain is largely driven by depressive symptoms. That is, although previous research has established that smoking is associated with increased pain [36,37], findings from this study suggest it is the relationship between smoking and depressive symptoms, and not the smoking per se that contributes to the self-report of greater pain. Therefore, increased attention to the role of depres-
Another aim of this study was to determine the relative impact of smoking in this setting. We found that the majority of patients seen in a tertiary care pain clinic (64%) were either current smokers (32%) or had a past history of smoking (32%). Previous studies have shown comparable smoking rates [14]; however, there is substantial variation in the literature [36]. Unlike the current study, most prior studies assessed smoking in specific groups of pain patients (eg, low back pain) or used epidemiological data. Because every new patient was sampled in the current study, 32% is likely an accurate reflection of smoking in this heterogeneous pain population. This rate is well beyond the smoking rate of 19% in the population’s home state of Michigan [8]. Smokers in the current study reported higher pain severity, greater pain interference, and more depressive symptoms compared to nonsmokers. Other studies have yielded similar results [27,36]. It appears that smoking is associated with a worse chronic pain phenotype, which may make smokers a more challenging group to treat from a clinical perspective with regards to their pain. Indeed, some studies have found worse long-term outcomes among smokers than nonsmokers following treatment [19,36]. Additionally, studies have shown an association between smoking and higher opioid doses [21,23], which could indicate smokers’
Ò
J. Goesling et al. / PAIN 153 (2012) 1749–1754
pain is harder to manage than nonsmokers’ pain. There are likely multiple reasons why smokers may be more difficult to treat, including the high prevalence of comorbid psychopathology. It is also probable, although not well understood, that nicotine interacts with the pharmacology of different drug treatments for pain [29], such as opioids. Future studies could help clarify if smokers are indeed more difficult to treat by including longitudinal assessments of posttreatment outcomes in both smokers and nonsmokers in outpatient and perioperative settings.
1753
measure of nicotine dependence, such as the Fagerstrom Test of Nicotine Dependence [19], should be used. Further consideration of the effect of dependence in this population is important, especially given the association between nicotine dependence and depression [4]. Lastly, although it is useful to know that smoking and pain are related, and that this relationship is mediated by depressive symptoms, there are likely other mechanisms that we did not consider in the current study. 4.5. Conclusions
4.3. Formers smokers resemble never smokers One unanswered question in the existing literature is ‘‘Does quitting smoking improve pain?’’ Researchers have turned to data on former smokers to answer this question. However, research comparing former smokers with current smokers (and nonsmokers) is limited and predominantly cross-sectional. Epidemiological studies have shown differences between former smokers and never smokers, with more pain reported in former smokers [27]. However, other studies [37] have found that former smokers reported pain at similar levels as nonsmokers. The finding from the current study that former smokers were indistinguishable from never smokers on pain and mood ratings helps strengthen the argument that quitting smoking may be beneficial. Additionally, the association between longer quit times and lower pain severity also points to a positive effect for quitting among former smokers. However, because these results rely on cross-sectional data, causal statements cannot be made, and further longitudinal and experimental research is needed in order to better understand how quitting smoking directly impacts pain outcomes. Given the high smoking rate of 32% in this treatment setting, providing smoking cessation assistance would likely be a valuable addition to treatment regardless of whether quitting actually reduces pain. However, it is unclear if current cessation interventions are effective among this population because little empirical attention has been devoted to evaluating cessation in outpatient pain clinics. The conclusions that have been drawn from this limited literature appear somewhat grim, with low rates of cessation among people in treatment for chronic pain [22]. Hooten and colleagues [24] published a seminal qualitative study assessing patient and physician attitudes towards smoking cessation. These authors found several patient-level barriers to cessation including using smoking as a way to manage pain-related emotional distress and as a distractor from pain. It is likely that the standard approach (eg, offering a single tobacco intervention) is not effective. Therefore, to advance the goal of providing effective smoking interventions to pain patients, future studies should include quantitative data collection on current cessation practices (eg, type of cessation offered) and outcomes (eg, do patients make quit attempts) and more in-depth studies on patients’ attitudes and motivation to quit. 4.4. Limitations This current study has several limitations. First, the data are cross-sectional, which limits our ability to assess for causality. This is especially relevant when considering the finding that former smokers were indistinguishable from never smokers. We were unable to assess when cessation occurred in relation to pain onset, and thus it is difficult to draw definitive conclusions about the effect of quitting smoking on pain. Second, the participants in the study were from an outpatient back and pain clinic and therefore may not be representative of chronic pain patients in the community or in different clinical settings. Another limitation was the measure of nicotine dependence (packs per day) used in this study. It was surprising that amount smoked was not related to any of the variables, especially depressive symptoms. Therefore, a better
In conclusion, one explanation for the association between smoking and pain points to depressive symptoms as a mediating factor. However, much remains unknown about this complex relationship, especially with regards to the impact that quitting smoking has on pain. The finding that longer quit duration was associated with less pain severity suggests that quitting may in fact be beneficial, but more rigorous experimental studies are needed. This study helps bring light onto several critical questions that remain to be answered in order to help advance the treatment of chronic pain patients. Conflicts of interest statement None of the authors have any relevant financial interest in this article or conflicts of interest. Acknowledgements This study was supported by the Department of Anesthesiology, University of Michigan Health System, Ann Arbor, MI, USA. We obtained permissions for use of the Brief Pain Inventory and Hospital Anxiety Depression Scale. The authors would like to thank Dr. Kevin K. Tremper, Professor and Chairman of the Department of Anesthesiology at the University of Michigan, Kevin Rakovitis and Taha Mahmood for their assistance with data entry, Baorong Shi for database development, and our support staff. References [1] Arnold LM, Keck Jr PE, Welge JA. Antidepressant treatment of fibromyalgia. A meta-analysis and review. Psychosomatics 2000;41:104–13. [2] Bair MJ, Robinson RL, Katon W, Kroenke K. Depression and pain comorbidity: a literature review. Arch Intern Med 2003;163:2433–45. [3] Baron RM, Kenny DA. The moderator-mediator variable distinction in social psychological research: conceptual, strategic, and statistical considerations. J Pers Soc Psychol 1986;51:1173–82. [4] Breslau N. Psychiatric comorbidity of smoking and nicotine dependence. Behav Genet 1995;25:95–101. [5] Brown RA, Kahler CW, Niaura R, Abrams DB, Sales SD, Ramsey SE, Goldstein MG, Burgess ES, Miller IW. Cognitive-behavioral treatment for depression in smoking cessation. J Consult Clin Psychol 2001;69:471–80. [6] Campbell LC, Clauw DJ, Keefe FJ. Persistent pain and depression: a biopsychosocial perspective. Biol Psychiatry 2003;54:399–409. [7] Centers for Disease Control and Prevention (CDC). Current cigarette smoking prevalence among working adults—United States, 2004–2011. MMWR Morb Mortal Wkly Rep 2011;60:1305–9. [8] Centers for Disease Control and Prevention (CDC). Tobacco control state highlights: Michigan. Available from: http://www.cdc.gov/tobacco/data_ statistics/state_data/state_highlights/2010/states/michigan/index.htm; [accessed 12.12.11]. [9] Cleeland CS, Ryan KM. Pain assessment: global use of the Brief Pain Inventory. Ann Acad Med Singapore 1994;23:129–38. [10] Covic T, Adamson B, Hough M. The impact of passive coping on rheumatoid arthritis pain. Rheumatology (Oxford) 2000;39:1027–30. [11] Dierker LC, Avenevoli S, Stolar M, Merikangas KR. Smoking and depression: an examination of mechanisms of comorbidity. Am J Psychiatry 2002;159: 947–53. [12] Eisenberger NI. The neural bases of social pain: evidence for shared representations with physical pain. Psychosom Med 2012;74:126–35. [13] Fishbain DA, Cutler R, Rosomoff HL, Rosomoff RS. Chronic pain-associated depression: antecedent or consequence of chronic pain? A review. Clin J Pain 1997;13:116–37.
1754
Ò
J. Goesling et al. / PAIN 153 (2012) 1749–1754
[14] Fishbain DA, Lewis JE, Cole B, Cutler RB, Rosomoff HL, Rosomoff RS. Variables associated with current smoking status in chronic pain patients. Pain Med 2007;8:301–11. [15] Gilron I, Bailey JM, Tu D, Holden RR, Jackson AC, Houlden RL. Nortriptyline and gabapentin, alone and in combination for neuropathic pain: a double-blind, randomised controlled crossover trial. Lancet 2009;374:1252–61. [16] Hall SM, Reus VI, Munoz RF, Sees KL, Humfleet G, Hartz DT, Frederick S, Triffleman E. Nortriptyline and cognitive-behavioral therapy in the treatment of cigarette smoking. Arch Gen Psychiatry 1998;55:683–90. [17] Hassett AL, Wasserman RA, Goesling J, Rakovitis K, Shi B, Brummett CM. The longitudinal assessment of pain outcomes in the clinical setting: Development of the ‘‘APOLO’’ electronic data capture system. RAPM, In Press. [18] Hauser W, Bernardy K, Uceyler N, Sommer C. Treatment of fibromyalgia syndrome with antidepressants: a meta-analysis. JAMA 2009;301: 198–209. [19] Heatherton TF, Kozlowski LT, Frecker RC, Fagerstrom KO. The Fagerstrom Test for Nicotine Dependence. a revision of the Fagerstrom Tolerance Questionnaire. Br J Addict 1991;86:1119–27. [20] Hellsing AL, Bryngelsson IL. Predictors of musculoskeletal pain in men: a twenty-year follow-up from examination at enlistment. Spine (Phila Pa 1976) 2000;25:3080–6. [21] Hooten WM, Shi Y, Gazelka HM, Warner DO. The effects of depression and smoking on pain severity and opioid use in patients with chronic pain. Pain 2011;152:223–9. [22] Hooten WM, Townsend CO, Bruce BK, Schmidt JE, Kerkvliet JL, Patten CA, Warner DO. Effects of smoking status on immediate treatment outcomes of multidisciplinary pain rehabilitation. Pain Med 2009;10:347–55. [23] Hooten WM, Townsend CO, Bruce BK, Warner DO. The effects of smoking status on opioid tapering among patients with chronic pain. Anesth Analg 2009;108:308–15. [24] Hooten WM, Vickers KS, Shi Y, Ebnet KL, Townsend CO, Patten CA, Warner DO. Smoking cessation and chronic pain: patient and pain medicine physician attitudes. Pain Pract 2011;11:552–63. [25] Jamison RN, Stetson BA, Parris WC. The relationship between cigarette smoking and chronic low back pain. Addict Behav 1991;16:103–10.
[26] Jensen MP, Turner LR, Turner JA, Romano JM. The use of multiple-item scales for pain intensity measurement in chronic pain patients. Pain 1996;67:35–40. [27] John U, Hanke M, Meyer C, Volzke H, Baumeister SE, Alte D. Tobacco smoking in relation to pain in a national general population survey. Prev Med 2006;43:477–81. [28] Keefe FJ, Brown GK, Wallston KA, Caldwell DS. Coping with rheumatoid arthritis pain: catastrophizing as a maladaptive strategy. Pain 1989;37:51–6. [29] Kroon LA. Drug interactions with smoking. Am J Health Syst Pharm 2007;64:1917–21. [30] Leboeuf-Yde C. Smoking and low back pain. A systematic literature review of 41 journal articles reporting 47 epidemiologic studies. Spine (Phila Pa 1976) 1999;24:1463–70. [31] MacKinnon DP, Lockwood CM, Hoffman JM, West SG, Sheets V. A comparison of methods to test mediation and other intervening variable effects. Psychol Methods 2002;7:83–104. [32] McCracken LM, Turk DC. Behavioral and cognitive-behavioral treatment for chronic pain: outcome, predictors of outcome, and treatment process. Spine (Phila Pa 1976) 2002;27:2564–73. [33] Murphy JM, Horton NJ, Monson RR, Laird NM, Sobol AM, Leighton AH. Cigarette smoking in relation to depression: historical trends from the Stirling County Study. Am J Psychiatry 2003;160:1663–9. [34] Phillips K, Clauw DJ. Central pain mechanisms in chronic pain states—maybe it is all in their head. Best Pract Res Clin Rheumatol 2011;25:141–54. [35] Shi Y, Weingarten TN, Mantilla CB, Hooten WM, Warner DO. Smoking and pain: pathophysiology and clinical implications. Anesthesiology 2010;113: 977–92. [36] Vogt MT, Hanscom B, Lauerman WC, Kang JD. Influence of smoking on the health status of spinal patients: the National Spine Network database. Spine (Phila Pa 1976) 2002;27:313–9. [37] Weingarten TN, Moeschler SM, Ptaszynski AE, Hooten WM, Beebe TJ, Warner DO. An assessment of the association between smoking status, pain intensity, and functional interference in patients with chronic pain. Pain Physician 2008;11:643–53. [38] Zigmond AS, Snaith RP. The hospital anxiety and depression scale. Acta Psychiatr Scand 1983;67:361–70.
PAIN 153 (2012) 1749–1754
www.elsevier.com/locate/pain
Cigarette smoking and pain: Depressive symptoms mediate smoking-related pain symptoms Jenna Goesling ⇑, Chad M. Brummett, Afton L. Hassett Department of Anesthesiology, University of Michigan, Ann Arbor, MI, USA
Sponsorships or competing interests that may be relevant to content are disclosed at the end of this article.
a r t i c l e
i n f o
Article history: Received 5 January 2012 Received in revised form 20 April 2012 Accepted 11 May 2012
Keywords: Smoking Chronic pain Depression Smoking cessation
a b s t r a c t Numerous studies have shown an association between smoking and pain, with smokers reporting more pain and worse functioning. However, little is known about factors that impact this complex relationship. This study investigated the association between smoking, pain, and depressive symptoms. Participants were new patients seen at a multidisciplinary pain clinic. All patients were mailed an intake packet of validated questionnaires as part of an ongoing research and clinical care initiative. Of the 497 patients evaluated, 426 had valid smoking data. Among these patients, 32.6% (n = 139) reported being current smokers, 31.7% (n = 135) were classified as former smokers, and 35.7% (n = 152) were never smokers. A multivariate analysis of covariance (smoking status, age, gender, education) revealed a main effect for pain severity (F = 7.36, P < 0.001), pain interference (F = 4.03, P = 0.001), and depressive symptoms (F = 7.87, P < 0.001). Current smokers demonstrated higher pain severity, pain interference, and depressive symptoms compared with former smokers and never smokers (P < 0.01 for all analyses), while there were no differences between the former-smoker and never-smoker groups. However, the effect of smoking on pain severity (P = 0.06) and pain interference (P = 0.22) was no longer significant after controlling for depressive symptoms in a mediation model. Additionally, among former smokers, longer quit duration was associated with less pain severity. In conclusion, smoking rates were high and smoking was associated with a worse chronic pain phenotype. Importantly, depressive symptoms emerged as a critical mediating factor in helping to explain the relationship between smoking and pain. Ó 2012 International Association for the Study of Pain. Published by Elsevier B.V. All rights reserved.
1. Introduction Smoking and chronic pain are recognized as 2 of the most challenging public health concerns. It is estimated that 20.6% of U.S. adults are current cigarette smokers [7]. The majority of studies with pain populations find that smoking rates are higher than the population rate [14,25]. For example, 54% of chronic low back pain patients reported being current smokers [25]. In addition to establishing smoking rates among pain patients, researchers have begun to characterize differences between smokers and nonsmokers with regards to pain-related complaints [23,36,37]. The majority of studies show that smoking is related to more pain [36,37]. In a population-based study, current smokers and former heavy smokers reported more pain locations and greater pain intensity compared to people who never smoked [27]. Similarly, data from
⇑ Corresponding author. Address: Back & Pain Center, Burlington Building 1, Suite 100, 325 E. Eisenhower Parkway, Ann Arbor, MI 48108, USA. Tel.: +1 734 232 1265; fax: +1 734 936 6585. E-mail address: [email protected] (J. Goesling).
the National Spine Institute found that in those with chronic back pain, smokers reported more severe pain than nonsmokers [36]. Several hypotheses have been advanced in attempting to unravel the nature of the association between smoking and pain. One possibility is that the aversive physiological effects of smoking cause or aggravate painful conditions [35]. However, it has been challenging to test this hypothesis because existing data are generally cross-sectional [30], and because there are many precipitating factors for pain. Thus, it is difficult to isolate causal relationships unless they are particularly strong. A meta-analysis of 41 studies on smoking and low back pain found that even when a positive association was found, effect sizes were small [30], suggesting that the evidence that smoking directly causes low back pain is weak. An alternative hypothesis is that smokers and people who have chronic pain share biological or environmental features. One variable of interest is depression, given its robust association with smoking [11,33] and pain [2,13]. Only one study has considered the multivariate associations between smoking, depression, and pain [21]. This study showed that smokers reported more pain but this association weakened when controlling for depression, suggesting that depression may in part explain why smoking is
0304-3959/$36.00 Ó 2012 International Association for the Study of Pain. Published by Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.pain.2012.05.014
1750
Ò
J. Goesling et al. / PAIN 153 (2012) 1749–1754
related to pain. Given the relatively weak evidence for a direct causal relationship between smoking and pain, further insight into comorbidities such as depression would be quite valuable. The present study was intended to address this issue in depth. The first aim of the current study was 2-fold: to establish smoking rates in a tertiary pain clinic and investigate the effects of smoking status on measures of pain and depressive symptoms. Specifically, it was hypothesized that smoking rates in our patient sample would be higher than the population rate, and smokers would report greater pain severity and pain interference as well as more symptoms of depression. The second aim was to test a mediation model specifically considering the role of depressive symptoms. It was predicted that depressive symptoms would partially account for the effect of smoking on pain measures. Finally, the third aim was to explore differences between former and current smokers. It was hypothesized that former smokers would have less pain and fewer symptoms of depression than current smokers. 2. Methods 2.1. Study setting and participants The Back & Pain Center (Department of Anesthesiology, University of Michigan Health System) is a multi-disciplinary outpatient pain medicine clinic where physicians treat a range of pain conditions and evaluate approximately 1400 new patients per year. Prior to the initial visit, all new patients are mailed a battery of validated questionnaires as part of an ongoing research and clinical care initiative to phenotype patients. A cover letter is included stating that some of the information from the questionnaires may be used for research, and a waiver of informed consent is obtained. It is stated in the cover letter that no identifying data will be disclosed should their information be used for research purposes. Patient packets are collected at the initial visit and entered into the Assessment of Pain Outcomes Longitudinal Electronic Data Capture system and exported for data analysis [17]. The collection of data and all subsequent analyses are approved by the Institutional Review Board at the University of Michigan (Ann Arbor, MI, USA). All new patients evaluated at the clinic from November 2010 through May 2011 who completed a new patient questionnaire were included in the initial analysis. Of the 497 patients evaluated to date, 71 could not be classified into 1 of the 3 smoking groups due to missing or incomplete data on the smoking items; therefore, they were excluded from the analyses. There were no other exclusion criteria. This resulted in 426 patients with complete smoking data who were included in the study. 2.2. Measures 2.2.1. Pain severity and pain interference The Brief Pain Inventory is a widely used, validated self-report scale that assesses the severity of pain and impact of pain of daily functioning [9]. Pain severity is assessed using a 4-item subscale that asks about pain at its worst, least, and average in the last week, and pain right now. For each item, patients rate their pain on a scale of 0-10 (0 = ‘‘No pain’’ and 10 = ‘‘Pain as bad as you can imagine’’). The average of the 4 scores can be used as a single composite measure of pain severity [26]. Pain interference is assessed using a 7-item subscale. Patients are asked to rate how much their pain interfered with different areas of their life (general activity, mood, walking ability, normal work, relations with others, sleep, and enjoyment of life) in the past week on a scale of 0-10 (0 = ‘‘Does not interfere’’ and 10 = ‘‘Completely interferes’’).
2.2.2. Depressive symptoms The Hospital Anxiety and Depression Scale is a brief and widely used instrument to measure psychological distress in general and medical populations [38]. It includes 2 7-item subscales that assess for anxiety and depressive symptoms. For the purpose of this study, we included only the depression subscale. The possible scores on this subscale range from 0 to 21. A score of 0-7 is considered within the normal range; a score of 8-10 is suggestive of the presence of depression; and a score of 11 or higher indicates that there is a high probability that depression is present [38]. 2.2.3. Smoking variables Patients were classified into 3 groups: (1) current smoker, (2) former smoker, and (3) never smoker. Smoking status was determined based on the question ‘‘Do you smoke?’’ (Yes = current smoker, No = current nonsmoker). Smoking history was assessed using the question, ‘‘Have you ever been a smoker?’’ Current nonsmokers who said, ‘‘yes’’ to this question were classified as former smokers. Former smokers provided additional information about how long it had been since they quit smoking, and number of years quit was calculated. If a patient said ‘‘no’’ to both questions (‘‘Do you smoke?’’ and ‘‘Have you ever been a smoker?’’), he/she was classified as a never smoker. Additionally, amount smoked was measured using current smokers’ response to the question, ‘‘How many packs per day?’’ 2.3. Data analysis Data were analyzed using SPSS (version 18; SPSS Inc, Chicago, IL, USA). Demographic characteristics (age, gender, education) were summarized in order to evaluate relevant characteristics of the sample. Mean and SD were reported for the continuous variable (age) and proportions were reported for categorical variables (gender and education), using analysis of variance and v2, respectively. Multivariate analysis of covariance was utilized to determine if smoking status was associated with pain severity, pain interference, and depressive symptoms. Due to differences in age, gender, and education across smoking groups, all 3 demographic variables were used as covariates in the above analyses. To test the hypothesis that depressive symptoms would mediate the effect of smoking on both pain severity and pain interference, we applied 2 convergent mediation procedures. The first, widely used procedure requires that 4 steps be satisfied to determine mediation [3]. First, a regression analysis was used to demonstrate an association between smoking status (current smoker: yes/no) and the outcome variables (pain). Second, another regression analysis was run to show an association between smoking status and the proposed mediator (depressive symptoms). Third, to assess the association between the proposed mediator (depressive symptoms) and the outcome variable (pain), a regression model was run. Lastly, pain was simultaneously regressed on both smoking status and depressive symptoms to determine if the effect of smoking on pain was reduced after adjusting for depressive symptoms. These 4 steps were done separately for pain severity and pain interference. An additional analysis was performed to address a potential limitation in the mediation results (one of the items on the Brief Pain Inventory pain interference scale measures mood, which is likely to overlap with depressive symptoms and could confound the model). We created a 6-item pain interference scale that excluded the mood item and reran the steps of the mediation analysis using this revised scale. The results did not change after the mood item was removed, so the model with the original validated 7-item scale is reported. For the second mediation procedure, we used a single, inclusive test of mediation (Sobel) as recommended by MacKinnon and colleagues [31]. The Sobel test
Ò
1751
J. Goesling et al. / PAIN 153 (2012) 1749–1754
[F(2, 423) = 20.94, P < 0.001]. Post hoc tests were performed and revealed that former smokers were more likely to be older than both current smokers and never smokers, Ps < 0.001. Current and never smokers did not differ on age, P = 0.24. Finally, the overall v2 was significant for education: v2 (2) = 17.57, P < 0.001. Current smokers differed from both former and never smokers (Ps < 0.0001) in that they were less likely to have graduated from college. Former and never smokers did not differ in regard to education (P = 0.11).
Table 1 Demographic differences between smokers, former smokers, and never smokers. Current smokers (n = 139)
Sex (male) Age College graduate (yes)
46.8%a 43.42a (12.02) 20.9%a
Nonsmokers Former smokers (n = 135)
Never smoked (n = 152)
51.5%a 55.44b (17.14) 35.1%b
32.9%b 46.7a (17.72) 44.5%b
Overall P-value
0.004 <0.0001 <0.0001
3.3. Effect of smoking on pain severity, pain interference, and depressive symptoms
Data presented as percentages or mean (SD). Values that do not share subscripts are significantly different at P < 0.05.
provides a conservative test of the indirect effect of the independent variable on the dependent variable via the mediator. It is often used as an additional test of mediation to complement the 4-step procedure described above. A significant Sobel test would suggest that the effect of smoking status on pain is largely driven by the intervening variable of depressive symptoms [31]. 3. Results 3.1. Smoking characteristics Among the 426 patients with valid smoking data, 32.6% (n = 139) reported being current smokers, 31.7% (n = 135) were classified as former smokers, and 35.7% (n = 152) were never smokers; 47.7% of smokers reported smoking a pack or more a day. Amount smoked per day was regressed on all 3 outcome variables in separate regression analyses. There was no effect for amount smoked on any of the outcome variables: pain severity (b = 0.06, t = 0.250, P = 0.80), pain interference (b = 0.05, t = 0.167, P = 0.868), or depressive symptoms (b = 0.62, t = 0.985, P = 0.326). The majority of former smokers (70.5%) reported having quit smoking over 5 years ago. Among the former smokers, number of years quit was regressed on all 3 outcome variables in separate regression analyses. There was no effect for number of years quit on pain interference (b = 0.005, t = 0.345, P = 0.730) or depressive symptoms (b = 0.021, t = 0.719, P = 0.473). However, there was an effect of number of years quit on pain severity (b = 0.028, t = 2.24, P = 0.027), with a shorter duration of being quit associated with higher pain severity. 3.2. Demographic characteristics The demographic characteristics of the participants are provided in Table 1. The overall v2 test for gender was significant: v2 (2) = 11.05, P = 0.004. Post hoc tests revealed there were no differences between current and former smokers with regard to gender (P = 0.43); however, patients who had never smoked were more likely to be female compared to both smoking groups (P < 0.001). There was also a significant main effect for age
Table 2 shows the effect of smoking on pain severity, pain interference, and depressive symptoms. A multivariate analysis of covariance with smoking status predicting all 3 outcome variables revealed a main effect for pain severity [F(5, 383) = 7.36, P < 0.000], pain interference [F(5, 383) = 4.03, P = 0.001], and depressive symptoms [F(5, 383) = 7.87, P < 0.001], with current smokers reporting greater pain intensity, more pain interference, and more symptoms of depression. Pairwise comparisons (Fisher’s least significant difference) revealed that these differences were significant for current smokers compared to former smokers (pain severity P = 0.008; pain interference P = 0.004; depressive symptoms P < 0.001) and never smokers (pain intensity P = 0.001; pain interference P = 0.001; depressive symptoms P < 0.001), but not between former smokers and never smokers (pain intensity P = 0.56; pain interference P = 0.65; depressive symptoms P = 0.78). 3.4. Mediation model First, to test the association between the primary predictor (smoking status: yes/no) and the first outcome variable (pain severity), pain severity was regressed on smoking status; b = 0.374, t = 3.52, P < 0.001. Second, to determine if there was an effect for smoking status on the proposed mediator, depressive symptoms was regressed on smoking status; b = 1.50, t = 5.58, P < 0.001. Next, the effect of the mediator, depressive symptoms, on pain severity was tested; b = 0.15, t = 8.53, P < 0.001. Finally, to test whether depressive symptoms mediated the relationship between smoking and pain, pain severity was regressed on smoking status, controlling for depressive symptoms; b = 0.197, t = 1.87, P = 0.06. The reduction in significance of smoking status when controlling for depressive symptoms showed that depressive symptoms mediated the effect of smoking on pain severity (Fig. 1). To further support this conclusion, we tested the indirect effect using a single (Sobel) test [31] and again found evidence for mediation, Sobel (Aroian) test = 4.44, P < 0.001. We tested mediation for the second outcome measure, pain interference, using the same approach. To test the association between the primary predictor (smoking status) and the second outcome variable (pain interference), pain interference was regressed on smoking status; b = 0.58, t = 4.32, P < 0.001. The second step, the effect of smoking status on the proposed mediator, was identical to
Table 2 Current smokers report more pain severity, pain interference, and depressive symptoms compared to former and never smokers. Current smokers
BPI pain severity BPI pain interference HADS depression subscale
6.90 (1.64) 7.51 (2.06) 11.20 (4.35)
Nonsmokers Former smokers
Never smoked
6.30 (1.88) 6.62 (2.24) 8.50 (4.45)
6.17 (1.87) 6.43 (2.48) 8.21 (4.60)
Current vs former P value
Current vs never P value
Former vs never P value
0.008 0.004 <0.001
0.001 0.001 <0.001
0.56 0.65 0.78
Data presented as mean (SD). BPI, Brief Pain Inventory: Pain severity (0-10) 0 = No pain, 10 = Pain as bad as you can imagine; Pain interference: (0-10) 0 = does not interfere, 10 = completely interferes; HADS, Hospital Anxiety and Depression scale (0-21); higher values indicate more depression.
Ò
1752
J. Goesling et al. / PAIN 153 (2012) 1749–1754
Depression
.15*
1.50*
Smoking
Pain Severity .20 (.37*)
Fig. 1. Path diagram demonstrating that depressive symptoms mediated the effect of smoking on pain severity. Reported statistics are unstandardized regression coefficients. ⁄P < 0.05.
Depression
.31*
1.50*
Smoking
.14 (.58*)
Pain Interference
Fig. 2. Path diagram demonstrating that depressive symptoms mediated the effect of smoking on pain interference. Reported statistics are unstandardized regression coefficients. ⁄P < 0.05.
that shown in the previous mediation analysis; b = 1.50, t = 5.58, P < 0.001. Next, the effect of depressive symptoms on pain interference was tested; b = 0.31, t = 16.63, P < 0.001. Finally, to test whether depressive symptoms mediated the relationship between smoking and pain, pain interference was regressed on smoking status, controlling for depressive symptoms; b = 0.139, t = 1.22, P = 0.22. The reduction in significance when controlling for depressive symptoms shows that depressive symptoms mediated the effect of smoking on pain interference (Fig. 2). Again, we tested the indirect effect using a single (Sobel) test and found evidence of mediation, Sobel (Aroian) test = 5.22, P < 0.001.
sion in smokers is an important precursor to the identification of the particular mechanisms that lead patients who smoke to report worse pain [10,28]. For example, cognitive models of depression propose that, because smokers are more likely to be depressed, smokers may make more negative appraisals as a result of believing they have less control over their pain [10,28]. Another possibility is that, because depression and pain are thought to share complex neurophysiological pathways [6,12,13,34], depressed smokers may experience more pain due to this interaction. Our mediation findings are consistent with Hooten and colleagues, who reported that the association between smoking and pain severity was reduced when depression was included in the statistical model [21]. In addition to testing such a model in a different treatment setting and on a more heterogeneous population of pain patients, the current study showed a relatively stronger mediating effect for depressive symptoms on pain severity and complete mediation on a pain outcome (pain interference) not previously tested. The consistency of results across these 2 studies highlights the generalizability of the mediation model for these variables. It follows that when exploring the association between smoking and pain, researchers should consider the role of depression in their participant samples. Interestingly, although many prior studies have found an association between smoking and pain [27,36], others have not [14]. One possible explanation for these conflicting findings may be patient-level differences on depression among various pain populations. For example, smokers with pain in the general population may be different from smokers being treated at pain clinics. In support of this possibility, research suggests that pain clinic patients tend to be more depressed than those in the community [2]. While we cannot make definitive causal statements based on these data, our results have tangible clinical value. The presence of depressive symptoms among current smokers being treated at the clinic was particularly striking, suggesting that smokers being treated for pain should also receive an evaluation for depression. Smokers with chronic pain exhibiting depression would likely benefit from cognitive-behavioral therapy and/or antidepressants in addition to current pain treatment [5,16] to help manage their depressive symptoms and because cognitive-behavioral therapy is also indicated for the treatment of chronic pain [32], and some classes of antidepressants have analgesic effects [1,15,18]. 4.2. Smoking rates and phenotypic characteristics of smokers
4. Discussion Given the impact of both cigarette smoking and chronic pain on the health and well-being of millions of Americans, understanding their relationship is both scientifically and clinically important. Our data suggest that smoking rates in patients being treated at an outpatient clinic were high, and smoking was associated with a worse chronic pain phenotype. Importantly, symptoms of depression emerged as a critical factor in helping to give a clearer interpretation of the relationship between smoking and pain. 4.1. Depressive symptoms mediated the relationship between smoking and pain A key finding from this study was that depressive symptoms mediated the relationships between smoking and pain severity as well as pain interference, suggesting that the link between smoking and pain is largely driven by depressive symptoms. That is, although previous research has established that smoking is associated with increased pain [36,37], findings from this study suggest it is the relationship between smoking and depressive symptoms, and not the smoking per se that contributes to the self-report of greater pain. Therefore, increased attention to the role of depres-
Another aim of this study was to determine the relative impact of smoking in this setting. We found that the majority of patients seen in a tertiary care pain clinic (64%) were either current smokers (32%) or had a past history of smoking (32%). Previous studies have shown comparable smoking rates [14]; however, there is substantial variation in the literature [36]. Unlike the current study, most prior studies assessed smoking in specific groups of pain patients (eg, low back pain) or used epidemiological data. Because every new patient was sampled in the current study, 32% is likely an accurate reflection of smoking in this heterogeneous pain population. This rate is well beyond the smoking rate of 19% in the population’s home state of Michigan [8]. Smokers in the current study reported higher pain severity, greater pain interference, and more depressive symptoms compared to nonsmokers. Other studies have yielded similar results [27,36]. It appears that smoking is associated with a worse chronic pain phenotype, which may make smokers a more challenging group to treat from a clinical perspective with regards to their pain. Indeed, some studies have found worse long-term outcomes among smokers than nonsmokers following treatment [19,36]. Additionally, studies have shown an association between smoking and higher opioid doses [21,23], which could indicate smokers’
Ò
J. Goesling et al. / PAIN 153 (2012) 1749–1754
pain is harder to manage than nonsmokers’ pain. There are likely multiple reasons why smokers may be more difficult to treat, including the high prevalence of comorbid psychopathology. It is also probable, although not well understood, that nicotine interacts with the pharmacology of different drug treatments for pain [29], such as opioids. Future studies could help clarify if smokers are indeed more difficult to treat by including longitudinal assessments of posttreatment outcomes in both smokers and nonsmokers in outpatient and perioperative settings.
1753
measure of nicotine dependence, such as the Fagerstrom Test of Nicotine Dependence [19], should be used. Further consideration of the effect of dependence in this population is important, especially given the association between nicotine dependence and depression [4]. Lastly, although it is useful to know that smoking and pain are related, and that this relationship is mediated by depressive symptoms, there are likely other mechanisms that we did not consider in the current study. 4.5. Conclusions
4.3. Formers smokers resemble never smokers One unanswered question in the existing literature is ‘‘Does quitting smoking improve pain?’’ Researchers have turned to data on former smokers to answer this question. However, research comparing former smokers with current smokers (and nonsmokers) is limited and predominantly cross-sectional. Epidemiological studies have shown differences between former smokers and never smokers, with more pain reported in former smokers [27]. However, other studies [37] have found that former smokers reported pain at similar levels as nonsmokers. The finding from the current study that former smokers were indistinguishable from never smokers on pain and mood ratings helps strengthen the argument that quitting smoking may be beneficial. Additionally, the association between longer quit times and lower pain severity also points to a positive effect for quitting among former smokers. However, because these results rely on cross-sectional data, causal statements cannot be made, and further longitudinal and experimental research is needed in order to better understand how quitting smoking directly impacts pain outcomes. Given the high smoking rate of 32% in this treatment setting, providing smoking cessation assistance would likely be a valuable addition to treatment regardless of whether quitting actually reduces pain. However, it is unclear if current cessation interventions are effective among this population because little empirical attention has been devoted to evaluating cessation in outpatient pain clinics. The conclusions that have been drawn from this limited literature appear somewhat grim, with low rates of cessation among people in treatment for chronic pain [22]. Hooten and colleagues [24] published a seminal qualitative study assessing patient and physician attitudes towards smoking cessation. These authors found several patient-level barriers to cessation including using smoking as a way to manage pain-related emotional distress and as a distractor from pain. It is likely that the standard approach (eg, offering a single tobacco intervention) is not effective. Therefore, to advance the goal of providing effective smoking interventions to pain patients, future studies should include quantitative data collection on current cessation practices (eg, type of cessation offered) and outcomes (eg, do patients make quit attempts) and more in-depth studies on patients’ attitudes and motivation to quit. 4.4. Limitations This current study has several limitations. First, the data are cross-sectional, which limits our ability to assess for causality. This is especially relevant when considering the finding that former smokers were indistinguishable from never smokers. We were unable to assess when cessation occurred in relation to pain onset, and thus it is difficult to draw definitive conclusions about the effect of quitting smoking on pain. Second, the participants in the study were from an outpatient back and pain clinic and therefore may not be representative of chronic pain patients in the community or in different clinical settings. Another limitation was the measure of nicotine dependence (packs per day) used in this study. It was surprising that amount smoked was not related to any of the variables, especially depressive symptoms. Therefore, a better
In conclusion, one explanation for the association between smoking and pain points to depressive symptoms as a mediating factor. However, much remains unknown about this complex relationship, especially with regards to the impact that quitting smoking has on pain. The finding that longer quit duration was associated with less pain severity suggests that quitting may in fact be beneficial, but more rigorous experimental studies are needed. This study helps bring light onto several critical questions that remain to be answered in order to help advance the treatment of chronic pain patients. Conflicts of interest statement None of the authors have any relevant financial interest in this article or conflicts of interest. Acknowledgements This study was supported by the Department of Anesthesiology, University of Michigan Health System, Ann Arbor, MI, USA. We obtained permissions for use of the Brief Pain Inventory and Hospital Anxiety Depression Scale. The authors would like to thank Dr. Kevin K. Tremper, Professor and Chairman of the Department of Anesthesiology at the University of Michigan, Kevin Rakovitis and Taha Mahmood for their assistance with data entry, Baorong Shi for database development, and our support staff. References [1] Arnold LM, Keck Jr PE, Welge JA. Antidepressant treatment of fibromyalgia. A meta-analysis and review. Psychosomatics 2000;41:104–13. [2] Bair MJ, Robinson RL, Katon W, Kroenke K. Depression and pain comorbidity: a literature review. Arch Intern Med 2003;163:2433–45. [3] Baron RM, Kenny DA. The moderator-mediator variable distinction in social psychological research: conceptual, strategic, and statistical considerations. J Pers Soc Psychol 1986;51:1173–82. [4] Breslau N. Psychiatric comorbidity of smoking and nicotine dependence. Behav Genet 1995;25:95–101. [5] Brown RA, Kahler CW, Niaura R, Abrams DB, Sales SD, Ramsey SE, Goldstein MG, Burgess ES, Miller IW. Cognitive-behavioral treatment for depression in smoking cessation. J Consult Clin Psychol 2001;69:471–80. [6] Campbell LC, Clauw DJ, Keefe FJ. Persistent pain and depression: a biopsychosocial perspective. Biol Psychiatry 2003;54:399–409. [7] Centers for Disease Control and Prevention (CDC). Current cigarette smoking prevalence among working adults—United States, 2004–2011. MMWR Morb Mortal Wkly Rep 2011;60:1305–9. [8] Centers for Disease Control and Prevention (CDC). Tobacco control state highlights: Michigan. Available from: http://www.cdc.gov/tobacco/data_ statistics/state_data/state_highlights/2010/states/michigan/index.htm; [accessed 12.12.11]. [9] Cleeland CS, Ryan KM. Pain assessment: global use of the Brief Pain Inventory. Ann Acad Med Singapore 1994;23:129–38. [10] Covic T, Adamson B, Hough M. The impact of passive coping on rheumatoid arthritis pain. Rheumatology (Oxford) 2000;39:1027–30. [11] Dierker LC, Avenevoli S, Stolar M, Merikangas KR. Smoking and depression: an examination of mechanisms of comorbidity. Am J Psychiatry 2002;159: 947–53. [12] Eisenberger NI. The neural bases of social pain: evidence for shared representations with physical pain. Psychosom Med 2012;74:126–35. [13] Fishbain DA, Cutler R, Rosomoff HL, Rosomoff RS. Chronic pain-associated depression: antecedent or consequence of chronic pain? A review. Clin J Pain 1997;13:116–37.
1754
Ò
J. Goesling et al. / PAIN 153 (2012) 1749–1754
[14] Fishbain DA, Lewis JE, Cole B, Cutler RB, Rosomoff HL, Rosomoff RS. Variables associated with current smoking status in chronic pain patients. Pain Med 2007;8:301–11. [15] Gilron I, Bailey JM, Tu D, Holden RR, Jackson AC, Houlden RL. Nortriptyline and gabapentin, alone and in combination for neuropathic pain: a double-blind, randomised controlled crossover trial. Lancet 2009;374:1252–61. [16] Hall SM, Reus VI, Munoz RF, Sees KL, Humfleet G, Hartz DT, Frederick S, Triffleman E. Nortriptyline and cognitive-behavioral therapy in the treatment of cigarette smoking. Arch Gen Psychiatry 1998;55:683–90. [17] Hassett AL, Wasserman RA, Goesling J, Rakovitis K, Shi B, Brummett CM. The longitudinal assessment of pain outcomes in the clinical setting: Development of the ‘‘APOLO’’ electronic data capture system. RAPM, In Press. [18] Hauser W, Bernardy K, Uceyler N, Sommer C. Treatment of fibromyalgia syndrome with antidepressants: a meta-analysis. JAMA 2009;301: 198–209. [19] Heatherton TF, Kozlowski LT, Frecker RC, Fagerstrom KO. The Fagerstrom Test for Nicotine Dependence. a revision of the Fagerstrom Tolerance Questionnaire. Br J Addict 1991;86:1119–27. [20] Hellsing AL, Bryngelsson IL. Predictors of musculoskeletal pain in men: a twenty-year follow-up from examination at enlistment. Spine (Phila Pa 1976) 2000;25:3080–6. [21] Hooten WM, Shi Y, Gazelka HM, Warner DO. The effects of depression and smoking on pain severity and opioid use in patients with chronic pain. Pain 2011;152:223–9. [22] Hooten WM, Townsend CO, Bruce BK, Schmidt JE, Kerkvliet JL, Patten CA, Warner DO. Effects of smoking status on immediate treatment outcomes of multidisciplinary pain rehabilitation. Pain Med 2009;10:347–55. [23] Hooten WM, Townsend CO, Bruce BK, Warner DO. The effects of smoking status on opioid tapering among patients with chronic pain. Anesth Analg 2009;108:308–15. [24] Hooten WM, Vickers KS, Shi Y, Ebnet KL, Townsend CO, Patten CA, Warner DO. Smoking cessation and chronic pain: patient and pain medicine physician attitudes. Pain Pract 2011;11:552–63. [25] Jamison RN, Stetson BA, Parris WC. The relationship between cigarette smoking and chronic low back pain. Addict Behav 1991;16:103–10.
[26] Jensen MP, Turner LR, Turner JA, Romano JM. The use of multiple-item scales for pain intensity measurement in chronic pain patients. Pain 1996;67:35–40. [27] John U, Hanke M, Meyer C, Volzke H, Baumeister SE, Alte D. Tobacco smoking in relation to pain in a national general population survey. Prev Med 2006;43:477–81. [28] Keefe FJ, Brown GK, Wallston KA, Caldwell DS. Coping with rheumatoid arthritis pain: catastrophizing as a maladaptive strategy. Pain 1989;37:51–6. [29] Kroon LA. Drug interactions with smoking. Am J Health Syst Pharm 2007;64:1917–21. [30] Leboeuf-Yde C. Smoking and low back pain. A systematic literature review of 41 journal articles reporting 47 epidemiologic studies. Spine (Phila Pa 1976) 1999;24:1463–70. [31] MacKinnon DP, Lockwood CM, Hoffman JM, West SG, Sheets V. A comparison of methods to test mediation and other intervening variable effects. Psychol Methods 2002;7:83–104. [32] McCracken LM, Turk DC. Behavioral and cognitive-behavioral treatment for chronic pain: outcome, predictors of outcome, and treatment process. Spine (Phila Pa 1976) 2002;27:2564–73. [33] Murphy JM, Horton NJ, Monson RR, Laird NM, Sobol AM, Leighton AH. Cigarette smoking in relation to depression: historical trends from the Stirling County Study. Am J Psychiatry 2003;160:1663–9. [34] Phillips K, Clauw DJ. Central pain mechanisms in chronic pain states—maybe it is all in their head. Best Pract Res Clin Rheumatol 2011;25:141–54. [35] Shi Y, Weingarten TN, Mantilla CB, Hooten WM, Warner DO. Smoking and pain: pathophysiology and clinical implications. Anesthesiology 2010;113: 977–92. [36] Vogt MT, Hanscom B, Lauerman WC, Kang JD. Influence of smoking on the health status of spinal patients: the National Spine Network database. Spine (Phila Pa 1976) 2002;27:313–9. [37] Weingarten TN, Moeschler SM, Ptaszynski AE, Hooten WM, Beebe TJ, Warner DO. An assessment of the association between smoking status, pain intensity, and functional interference in patients with chronic pain. Pain Physician 2008;11:643–53. [38] Zigmond AS, Snaith RP. The hospital anxiety and depression scale. Acta Psychiatr Scand 1983;67:361–70.