OUP user menu

Does Smoking Status Affect Multidisciplinary Pain Facility Treatment Outcome?

David A. Fishbain MD, FAPA, John E. Lewis PhD, Robert Cutler PhD, Brandly Cole PsyD, R. Steele Rosomoff BSN, MBA, Hubert L. Rosomoff MD, DMedSc, FAAPM
DOI: http://dx.doi.org/10.1111/j.1526-4637.2007.00306.x 1081-1090 First published online: 1 November 2008

ABSTRACT

Objectives. Smoking may be a major problem in chronic low back pain (LBP) patients. The goal of this study was to determine whether smoking status affected multidisciplinary pain facility treatment outcome.

Design. As part of a grant study, chronic LBP patients identified themselves as either current smokers (N = 81) or current nonsmokers (N = 140), and were compared by chi-square for employment status at 1, 6, 12, and 24 months after multidisciplinary pain facility treatment. Smokers who were unemployed at each time interval were then compared with employed smokers for a large number of assessment scales and clinical variables of interest by chi-square or Student's t-test. The significant independent variables from these analyses were then utilized in a logistic regression to determine predictors for smoker nonemployment.

Setting. Pain facility.

Results. Current smokers were less likely to be employed at each follow-up time point. Pain levels over the previous 24 hours predicted employment status for current smokers at 1-, 12-, and 24-month follow-up, while worker compensation status predicted employment status at 6 months.

Conclusions. Current smoking status appears to be associated with poorer treatment outcome after multidisciplinary pain facility treatment. Return to work within smokers is predicted by pain and worker compensation status. Pain facilities should target current smokers with significant perceived pain for close treatment monitoring in an attempt to improve treatment outcome.

  • Smoking
  • Pain Facility Treatment Outcome
  • Chronic Low Back Pain
  • Nicotine
  • Pain

Introduction

It has been reported that the percentage of smokers within chronic low back pain (LBP) patients is higher than the national smoking percentage [1]. Additionally, smoking may be a risk factor for chronic LBP [2–6]. At issue, which then has some relevance to pain treatment outcome, is whether smokers with chronic LBP are less likely to achieve a positive outcome from pain treatment than nonsmokers with chronic LBP.

The above question has been addressed indirectly by four previous studies. These studies were consistent in indicating that smokers are more likely to be disabled by their LBP, but unfortunately these studies did not address the issue of treatment outcome. In the first prospective study, Lanier and Stockton [7] reported that smokers who developed acute mechanical LBP demonstrated greater long-term disability than nonsmokers. In a second study, Krouse-Wood et al. [8] demonstrated that smoking status was one of the predictors that could explain physician-determined fit-for-work status for patients reporting complaints of LBP. Oleske et al. [9], in the third prospective study, reported that auto workers with work-related LBP demonstrated levels of LBP disability according to their levels of cigarette smoking. In the fourth study, Oleske et al. [10] demonstrated that current cigarette smoking was a factor associated with higher disability levels in workers diagnosed with work-related LBP disorders.

If smoking status is related to higher levels of disability in newly injured LBP patients, then treatment outcome with smokers should be worse than that for nonsmokers with LBP. The authors were only able to find two studies which addressed this issue. In the first study, Vogt [11] reported that smokers demonstrated little improvement in physical and mental health after spinal surgery while nonsmokers showed progressive improvement. In the second study, McGeary et al. [12] reported that, after completion of a functional restoration program, smokers at 1-year follow-up did not differ from nonsmokers on work status, but did differ on work retention. Thus, in terms of pain treatment outcome, there are limited data as to whether smoking impacts on this issue. In addition, the limited data appear to be contradictory. As such, and in order to shed light on this problem, the research group at the Rosomoff Pain Center performed the study outlined below.

Methods

Between March 1991 and March 1993, over 1,000 consecutive chronic pain patient (CPP) admissions to the University of Miami Comprehensive Pain Center were screened for possible selection for a National Institute on Disability and Rehabilitation Research Grant Study. Because this grant study dealt with prediction of return to employment postpain facility treatment, each selected CPP received a detailed assessment at admission and follow-up posttreatment. CPP inclusion criteria for the grant study were: (a) candidate for employment posttreatment (not a student, not a housewife by profession, not retired, not receiving social security disability, or not accepted for social security disability); (b) age range 19–62; (c) chronic LBP (more than 6 months' duration) as a presenting problem; (d) able to read English; (e) not requiring surgery for LBP at the time of admission; and (f) willing to sign the informed consent for participation in this grant study.

After signing the consent form, the chronic LBP patient completed a series of baseline demographic questionnaires and psychological inventories. These included: basic demographic information, including a smoking question; visual analog scale (VAS) scale (0–100); pain disability questionnaire; sickness impact profile; functional assessment total score; Beck Depression Inventory; and the State–Trait Anxiety Inventory. The smoking question was worded: “How many cigarettes do you currently smoke per day?” (circle) none, less than ¼ pack, ¼ to ½ pack, ½ to ¾ pack, ¾ to 1 pack, 1 to 1½ packs, 1½ to 2 packs, and greater than 2 packs. In addition, each chronic LBP patient received a psychiatric diagnostic interview by a senior psychiatrist utilizing Diagnostic Statistical Manual, 3rd Edition Revised (DSM-III-R) flow sheets and criteria.

All chronic LBP patients entering the grant study then received 1 month of multidisciplinary (multimodal) pain treatment. This consisted of physical therapy, occupational therapy, rehabilitation counseling, biofeedback, relaxation, and pharmacological management for pain and psychiatric comorbidity. All opioids were tapered during the treatment period. No chronic LBP patient entering the grant study was considered by the neurosurgeon and physiatrist after the initial evaluation to be a candidate for surgical treatment and/or other forms of invasive treatment (e.g., blocks), as most of them had already been failures in these treatments.

Those chronic LBP patients entering the grant study were followed up after this 1 month of multimodal treatment at 1, 6, 12, and 24 months. At each follow-up time point, work status of the chronic LBP patient was determined and recorded. Follow-up evaluation for work status consisted of a four-step procedure, in which if one method failed, the next would be used. These steps were:

  1. Mailed questionnaire;

  2. Intensive and continuing telephoning nights and weekends using the same questionnaire;

  3. Review of follow-up medical records for work status; and

  4. Contact with insurance carrier and referring physician for work status.

For the purposes of the grant study, chronic LBP patients were placed into the following vocational categories at follow-up evaluation through this four-step procedure: employed full-time or part-time, employed with restrictions or without restrictions, unemployed, and the exact type of employment. The possible unemployed categories were: looking for work without restrictions, getting retraining for a new job, waiting for retraining for a new job, looking for work with restrictions, student by profession, attending vocational or training school, housewife by profession, retired and receiving social security, retired and not receiving social security, applied for social security, accepted for or receiving social security, able to work but cannot find a job, pain condition too severe to work, poor health other than pain prevents working, no desire to find employment, and other miscellaneous reasons associated with unemployment. Smoking status was not determined at follow-up time points.

Additional methodological comments about determination of work status are as follows. We considered that any chronic LBP patient was a candidate for return to some form of work unless he/she was excluded from working by our exclusion criteria. During treatment, no one who entered the study converted to disability status. One hundred percent of the chronic LBP patients entering the study were unemployed before admission to the pain center. It was expected, and attempts were made to accomplish this (unless the chronic LBP patient refused), that the chronic LBP patient would return to his/her preinjury type of work. For the purposes of this study, however, chronic LBP patients were considered as having returned to work whether or not they returned to their previous type of work. Other articles have been published as a result of this grant study [13–25].

Data Analysis

Data were analyzed using the Statistical Package for the Social Sciences software. Frequency and descriptive statistics were calculated to check all relevant characteristics of the data. Patients classified themselves as either current smokers (N = 81) or current nonsmokers (N = 140), and gave the number of cigarettes smoked per day according to packs. This classification was used as the dependent variable for analyses to determine differences on the instrument scores and scales, and whether or not the patient had returned to part- or full-time work at 1-, 6-, 12-, and 24-month follow-up. Smokers were divided into those who had returned to work and those who had not, to assess differences on the study assessments and scales and other background and clinical variables of interest. The Student's t-test or chi-square was utilized to detect differences in continuous and categorical variables, respectively. Logistic regression was used to determine the predictors of whether or not a smoker was employed at each follow-up by inputting significant independent variables (P < 0.01) from prior analyses. The alpha level used for all analyses was set at 0.01.

Only current DSM-III-R diagnoses were utilized in the analyses. In addition, for the purposes of the analyses, drug-abuse and drug-dependence diagnoses for similar drugs (e.g., alcohol, cannabis, etc.) were collapsed into one category, such as current alcohol abuse/dependence, current cannabinoid abuse/dependence, etc.

Results

This sample was composed of 58% (N = 128) being male and 42% (N = 93) being female, with a mean age of 41.1 years (SD = 10.0, range = 19, 62). The ethnic distribution was 164 (74.2%) white, non-Hispanic; 18 (8.1%) black, non-Hispanic; 19 (8.6%) Hispanic; and 20 (9.1%) of unknown racial origin. The sample of smokers consisted of 63% (N = 51) male and 37% (N = 30) female participants. The average age of the smokers was 40.6 years (SD = 9.7; range = 19, 59). The ethnic distribution was: 85% (N = 69) white, non-Hispanic; 7.5% (N = 6) black, non-Hispanic; and 7.5% (N = 6) of unknown racial origin. None of the Hispanics were smokers in our sample. The frequency distribution for number of packs of cigarettes smoked for the sample is presented in Table 1. Approximately 37% of the subjects entering the grant study classified themselves as smokers, while 15.4% were heavy smokers (more than 1 pack per day).

View this table:
Table 1

Frequency distribution for the number of packs of cigarettes smoked per day for the sample (N = 221)

Frequency%
None (N)140 63.3
Greater than 2 packs  3  1.4
1½ to 2 packs  6  2.7
1 to 1½ packs 25 11.3
¾ to 1 pack 18  8.1
½ to ¾ pack 11  5.0
¼ to ½ pack 9  4.1
Less than ¼ pack 9  4.1
Total221100.0

Smokers were compared with nonsmokers on several demographic variables. Smokers were not significantly older (M = 41.4 years, SD = 10.2) than nonsmokers (M = 40.6 years, SD = 9.7; t[218] = 58, P = 57). The proportion of male participants (63%) was not significantly different among smokers than among nonsmokers (55%; χ2 = 1.3 [1], P = 0.25). Forty-two percent of the non-Hispanic whites, 33% of the non-Hispanic blacks, none of the Hispanics, and 31% of other races were smokers (χ2 = 13.3 [3], P < 0.01). An equal proportion of smokers and nonsmokers were married, single, separated, divorced, or widowed (χ2 = 2.2 [5], P = 0.82). Fifty-five percent of the sample who completed high school or less were smokers, whereas 14% of the patients who had at least completed 2 years of college were smokers (χ2 = 32.1 [7], P < 0.001). All chronic LBP patients entering this grant study, whether smokers or nonsmokers, were unemployed at time of entrance.

Success for contact at follow-up time points for the sample group was (beginning with N = 221): at 1 month, 215 or 97.2%; at 6 months, 93.6%; at 12 months, 85.9%; and at 24 months, 47.9%. Overall, for all the possible time points, CPPs were contacted and followed up successfully for 81.2% of the time points. The major difference that predicted successful contact was that of worker compensation status [22].

Comparisons Between Smokers and Nonsmokers for Employment Status at Follow-Up (Table 2)

View this table:
Table 2

Comparisons between smokers and nonsmokers and employment status at follow-up

VariableCategory1-Month Employment6-Month Employment12-Month Employment24-Month Employment
SmokerYes

No
Y = 27 (35.5%)
N = 49 (64.5%)
Y = 74 (53.2%)
N = 65 (46.8%)
Y = 23 (32.4%)
N = 48 (67.6%)
Y = 76 (55.9%)
N = 60 (44.1%)
Y = 15 (23.8%)
N = 48 (76.2%)
Y = 78 (61.4%)
N = 49 (38.6%
Y = 9 (23.1%)
N = 30 (76.9%)
Y = 35 (52.2%)
N = 32 (47.8%)
χ2 (df)6.2 (1)*10.3 (1)*23.8 (1)*8.6 (1)*

Results are displayed as percentages within each category. Significantly higher rates of unemployment were found for smokers compared with nonsmokers at each time point. At 1-month follow-up, 36% of smokers were employed, compared with 53% of nonsmokers (χ2 = 6.2 [1], P < 0.01). Thirty-two percent of smokers were employed at 6-month follow-up, while 56% of nonsmokers were working (χ2 = 10.3 [1], P < 0.01). Less than 24% of smokers were working at 12-month follow-up, but 61% of the nonsmokers were employed either full- or part-time (χ2 = 23.8 [1], P < 0.01). Twenty-three percent of smokers were employed at 24-month follow-up, compared with 52% of the nonsmokers (χ2 = 8.6 [1], P < 0.01).

Further Analyses and Results Within Smokers

We also wished to determine whether some variables discriminate between smokers who do and do not return to work. As we had a relatively small sample size of smokers (N = 81), we calculated power based on 70% of smokers not returning to work. This calculation resulted in 0.81 power, indicating that we had enough power to conduct the analyses below.

The frequency for the following demographic and clinical variables was compared between smokers who do and do not return to work at 1-, 6-, 12-, and 24-month follow-up. These variables were: age, gender, marital status, worker compensation status, severely fatigued or not, Beck depression score, functional assessment total score, sickness impact profile score, VAS average pain over the last 24 hours, VAS-level pain considered intolerable, VAS-level pain over the last 24 hours, VAS-level pain considered intolerable, VAS-level pain to take medications, VAS-level pain considered disabling, pain disability index score, State Anxiety total score, Trait Anxiety total score, opioid abuse/dependence, inhalant abuse/dependence, psychoactive abuse/dependence, cocaine abuse/dependence, sedative abuse/dependence, major depressive disorder, dysthymic disorder, adjustment disorder with depressed mood, panic disorder, generalized anxiety disorder, adjustment disorder with anxious mood, and conversion disorder.

We found that the VAS average pain score over the last 24 hours consistently discriminated between those smokers who were employed and those who were not (Table 3). Those smokers who were not employed scored higher than the smokers who were employed at each follow-up assessment. The pain disability index score was higher for smokers who were not employed at 1-month follow-up, but was nonsignificant at subsequent time points (Table 3). At 1-month follow-up, a higher percentage of smokers who were not working were diagnosed with conversion disorder (32.7%), compared with those who were working (7.4%; χ2 = 6.2 [1], P < 0.01) (Table 4). At 6-month follow-up, 87.5% of those smokers not employed were worker compensation patients vs 56.5% for the smokers who were working (χ2 = 8.5 [1], P < 0.01; Table 4). None of the other measures were statistically different between smokers who were and who were not employed at the follow-up assessments.

View this table:
Table 3

Comparisons among smokers who do and do not return to work for continuous demographic and clinical characteristics

Did Return to Work, Mean (SD)Did Not Return to Work, Mean (SD)  t (df)
1-month employmentVAS pain score in previous 24 hours 6.2 (2.43) 8.0 (1.95)3.6 (71)*
Pain disability index39.7 (15.8)48.0 (11.5)2.6 (67)*
6-month employmentVAS pain score in previous 24 hours 6.1 (2.72) 7.8 (1.90)3.2 (67)*
12-month employmentVAS pain score in previous 24 hours 5.2 (2.31) 7.8 (2.00)4.3 (60)*
24-month employmentVAS pain score in previous 24 hours 5.4 (2.84) 8.2 (1.73)3.6 (36)**
  • * P < 0.01;

  • ** P < 0.001.

  • VAS = visual analog scale.

View this table:
Table 4

Comparisons among smokers who do and do not return to work for categorical demographic and clinical characteristics

Did Return to WorkDid Not Return to Workχ2 (df)
1-month employmentConversion disorderY = 2 (7.4%)
N = 25 (92.6%)
Y = 16 (32.7%)
N = 33 (67.3%)
6.2 (1)*
6-month employmentWorker compensationY = 13 (56.5%)
N = 10 (43.5%)
Y = 42 (87.5%)
N = 6 (12.5%)
8.5 (1)*

Logistic Regression for Significant Demographic and Clinical Characteristics as the Independent Variables with Employment as the Dependent Variable Among Smokers

Next, a logistic regression was conducted for each follow-up assessment with the significant variables from prior analyses as the independents and employment as the dependent variable, with employed coded as “1” and unemployed coded as “0.”Table 5 includes the regression coefficient, model χ2 and significance, Nagelkerke R2, Wald statistic, odds ratio, and 95% confidence intervals for the odds ratio. The overall chi-squares were significant at each follow-up assessment, and the models classified a higher proportion of patients correctly at each successive time point. While the VAS pain score in the previous 24 hours was significant at the 1-, 12-, and 24-month follow-up time points and worker compensation was significant at 6-month follow-up, the odds ratios of 0.71, 0.32, 0.82, and 0.82, respectively, show little change in the likelihood of employment on the basis of a one-unit change in either variable. However, the probability of unemployment increased by a multiplicative factor of 3.15 with the use of worker compensation at 6-month follow-up. Other variables were insignificant and were not retained in the final model.

View this table:
Table 5

Final model logistic regression results for smokers with employment as the dependent variable

χ2 (df)% of Cases Predicted Correctly by the ModelNagelkerke R2VariableBWaldSigOdds Ratio95% Cl for Odds Ratio
LowerUpper
1-month employment 8.7 (1)*71.00.16VAS pain score in previous 24 hours−0.34 7.480.010.710.560.91
6-month employment15.3 (1)*74.00.26Worker compensation−1.1513.10.0010.320.1700.59
12-month employment26.2 (1)*76.00.46VAS pain score in previous 24 hours−0.2019.10.0010.820.750.90
24-month employment16.8 (1)*76.00.48VAS pain score in previous 24 hours−0.2012.10.0010.820.730.92
  • * P < 0.01 for the overall model chi-square.

  • Sig = significance; VAS = visual analog scale.

Logistic Regression for Significant Demographic and Clinical Characteristics as the Independent Variables with Employment as the Dependent Variable Among Smokers who Smoke Either up to 1 Pack per Day or More than 1 Pack per Day

As noted above, the previous analyses were performed with the group of smokers who do and do not return to work at each follow-up interval. Of interest, however, was whether the results of the above analyses differed if the smoking group was divided into those smokers who smoked up to 1 pack per day (N = 47) and those who smoked more than 1 pack per day (N = 34). In similar sequential fashion to the previous analyses, chi-squares were conducted for categorical variables, and t-tests were conducted for continuous variables to determine which independent variables met statistical criteria (P ≤ 0.01) to build logistic regression models at 1-, 6-, 12-, and 24-month follow-up (Table 6). We were not able to conduct a regression for the 1-month employment variable for smokers who smoked more than 1 pack per day. All model chi-squared values were statistically significant, except for the 12-month employment variable for smokers who smoked more than 1 pack per day, and the models classified a higher proportion of patients correctly at each successive step in the model. Table 6 includes the chi-squared values and significance, % of cases predicted correctly by the model, and the Nagelkerke R2 at each step in the model, and the Wald statistic and significance, odds ratio, and 95% confidence intervals for the odds ratio for the final step in the model. While the VAS pain score in the previous 24 hours was significant in all of the time points except for 6-month employment for smokers who smoked up to 1 pack, 1-, 12-, and 24-month follow-up equations, the odds ratios of just 1 show little change in the likelihood of employment on the basis of a one-unit change in this variable. However, the probability of employment increased by a multiplicative factor of 2, with a higher level of education for 1-month employment among smokers who smoked up to 1 pack per day, and by a multiplicative factor of 4.2, with more education for 6-month employment among smokers who smoked more than 1 pack per day.

View this table:
Table 6

Final model logistic regression results for smokers by amount smoked with employment as the dependent variable

χ2 (df)% of Cases Predicted Correctly by the ModelNagelkerke R2VariableBWald P valueOdds Ratio95% Cl for Odds Ratio
LowerUpper
Up to 1 pack smokers
1-month employment 8.00 (1)**67.50.24VAS pain score in previous 24 hours−0.45 9.920.0020.6390.483 0.844
 9.16 (1)**75.00.22Education 0.70 6.720.012.011.19 3.41
6-month employment11.6 (1)**74.40.34Worker compensation VAS pain score in previous 24 hoursEducation−1.39 9.23  0.01  1.120.002 0.93 0.280.250.102 0.612
12-month employment23.2 (1)**85.70.65VAS pain score in previous 24 hours−0.2813.30.0010.760.65 0.88
24-month employment 7.9 (1)**76.20.42VAS pain score in previous 24 hours−0.18 6.040.0140.840.72 0.96
More than 1 pack smokers
1-month employmentVAS pain score in previous 24 hoursEducation 2.30  0.000.13 1.0
6-month employment43 (1)*73.30.18VAS pain score in previous 24 hours−0.484 4.540.0330.620.40 0.96
12.9 (1)**83.30.40Education 1.43 7.430.0064.191.5011.73
 5.5 (1)*86.70.13Worker compensation−2.78 3.090.0790.060.003 1.377
12-month employment 5.4 (1)*63.00.24VAS pain score in previous 24 hours−0.13 4.70.0310.880.78 0.99
24-month employment 9.02 (1)**76.50.55VAS pain score in previous 24 hours−0.23 5.950.0150.800.67 0.96
  • * P < 0.05;

  • ** P < 0.01.

  • VAS = visual analog scale.

There were no differences in the predictor variables between smokers who smoked up to 1 pack per day and those who smoked more than 1 pack per day for the 12- and 24-month follow-up time points. This comparison could not be made for the 1-month time point. For the 6-month time point for smokers who smoked up to 1 pack per day, only one variable entered the model: worker compensation status. For the smokers who smoked more than 1 pack per day, two additional variables entered the model: VAS pain score in previous 24 hours and education.

Discussion

The results of this study indicate that CPPs who smoke are less likely to be employed at all follow-up time points than their counterparts. However, this finding implies an association and not necessarily causality. As noted in the Introduction, this finding is supported by the majority of previous studies [7–11] that have indirectly [7–10] and directly [11] addressed this issue. Contrary findings were only generated in one study [12]. The results of our study have also demonstrated that the work status of smoking chronic LBP patients can be predicted by only two variables: pain for the 1st, 6th, 12th, and 24th month follow-up interval, and worker compensation status for the 6-month follow-up interval. If this smoking group is broken down into those who smoke 1 pack or less per day and those who smoke more than 1 pack per day, the predictor variables essentially remain the same (pain, worker compensation status), but education status is added. These factors are interesting because, in a previous study, Fishbain et al. [13] demonstrated that smoking status in CPPs was not predicted by pain levels and, contrary to expectations, was also not predicted by presence of mood disorders, anxiety syndromes, illicit drug-use disorders (cocaine, cannabinoids, etc.), fatigue, and insomnia. Thus, it appears that within smokers, there are two groups: those characterized by less pain and who are employed at 1, 12, and 24 months; and a second group characterized by more pain and unemployment at 1, 12, and 24 months. As these results are from a logistic regression, one can conclude that the reason why this last group does not go back to work at 1, 12, and 24 months is pain. It is also interesting to note that worker compensation status was associated with unemployment at 6 months, but not at other time points. This is not surprising as there is a wide body of literature on the association of worker compensation status and poor treatment outcome [26,14–16,27] as measured by return to work. Thus, our results are supported and support these previous studies. At issue, however, is why this variable was only predictive at the 6-month time point? The reasons for this are not absolutely clear, but it is likely that this relates to CPP movement in and out of work after treatment/rehabilitation. This has been demonstrated in an earlier study by our group with this same data set [17].

If nonemployed chronic LBP smokers do have more pain, what does the literature say about by which mechanism this occurs? There is some evidence [28] that the vasoconstrictive properties of nicotine may cause vasoconstriction around the discs, thus impairing circulation leading to lowered nutrition. This in turn would lead to premature disc degeneration and more pain. In addition, the association of smoking and chronic LBP is strongest among persons who suffer from respiratory diseases [29]. Smoking influences the immune system. Thus, smokers have been shown to have T-cell abnormalities, reduced levels of natural killer cells and antigen presenting cells [30]. In addition, smokers have higher levels of androgens, which are immunosuppressive [30]. Immunosuppression predisposes to respiratory illness, which in turn may lead to protracted coughing. This may lead to mechanical stress on spinal structures [31]. This then again may lead to premature disc degeneration and pain.

It is also to be noted here that the results of this study indicate that depression was not related to employment status in smokers. This finding is interesting because there are some psychopharmacological data on nicotine. It appears that chronic administration of nicotine decreases the concentration and biosynthesis of serotonin in rats, thus impairing serotonergic function, which may trigger depression [32]. At the same time, there are substances in smoke which inhibit mono amine oxidase (MAO) and therefore have an antidepressant effect. In addition, nicotine is an agonist at the nicotine receptor and thus increases the firing of the locus coeruleus (LC), stimulating the release of norepinephrine from LC neurons [32]. This activity is therefore antidepressant in nature. It has therefore been proposed that smoking may correct neurochemical deficits associated with depression and that smokers smoke to correct this deficit [32]. If this last biochemical observation is true, clinically one would expect that smokers would not be depressed because this deficit is corrected by smoking. Our results are then in accord with this last hypothesis.

Early multidisciplinary pain treatment outcome studies have demonstrated that educational level is a predictor of treatment outcome [27]. There is also evidence that, within the general population, a greater percentage of blue-collar workers smoke than white-collar workers. Unfortunately, whether more blue-collar workers smoke than white-collar workers has not been examined for chronic LBP patients, but it is likely that the same observation applies to this universe. It is also unknown whether the universe of chronic LBP patients contains more blue-collar workers vs white-collar workers. Nevertheless, it is the observation of physicians who work in this area that, within chronic LBP patients, blue-collar workers represent a substantial percentage. This would be expected as these patients are more likely to be involved in a job that requires lifting, which is generally observed to be associated with injury to the low back. Finally, blue-collar workers are more likely to have lower educational status than white-collar workers. Our results indicate that, within the universe of smokers, higher educations status did predict return to work at 6 months. Thus, our data are supported by the above discussion, and the above discussion could explain the reason for the education finding.

Are there limitations or confounders to this study's results? The first possible confounder is that of selecting chronic LBP pain patients who are potential return-to-work candidates. This type of selection could have increased the numbers of smokers within the total patient sample. The smoking patient could potentially be more likely to be a blue-collar worker with lower education status, which has been shown to be a predictor of nonreturn to work after multidisciplinary treatment [27]. As such, this could have affected our results on smokers being more likely not to return to work. We reanalyzed our data with potential confounders entered into the logistic regression models, but our results were not altered with utilizing such an approach. The second potential confounder was that of selecting chronic LBP patients who chose to enter the multidisciplinary pain center and chose to proceed with opioid detoxification if necessary. It is unclear how this aspect could have influenced the results. However, this confounder would alert us to the possibility that these results may not be generalizable to all chronic LBP patient samples. The third potential confounder is that of the large number of statistical tests that were conducted potentially generating a significant variable by chance alone. This is a possibility. However, to offset this potential confounder, only the variables significant at P ≤ 0.01 were selected for further analysis. This strategy we believe minimized the possibility of this confounder impacting significantly on our results. Finally, there are limitations in applying the results of this study to other chronic LBP populations. This is because other chronic LBP populations may differ in their percentages of blue-collar workers and candidates for return to work. In addition, these data are 15 years old. During this period, the prevalence of smoking has dropped, but may not have dropped proportionally in all subgroups; that is, white-collar workers might have been more likely to quit. This would increase the proportions of smokers who are blue collar within chronic LBP groups. As such, these results may not apply to current LBP groups. These data may therefore need to be replicated in current LBP groups.

What are the clinical implications of this study? First, pain clinicians should now be aware that treatment outcome with chronic LBP smokers could be worse than for chronic LBP nonsmokers. There may be a subgroup of smoking chronic LBP patients who have significant perceived pain and whose outcome will be dependent on the amelioration of this pain. Thus, this group should be targeted for treatments that may improve pain and thus improve outcome. Second, this study raises the issue of whether pain facility treatment should involve a smoking cessation program in order to improve outcome. This last question can only be answered by a future study where a smoking cessation program is included into a treatment package for chronic LBP patients. Outcome results for such a package could then be compared with a package without such a program. We have had clinical experience with trying to add a smoking cessation program to a chronic LBP treatment package. We found that such a program increased the difficulties in dealing with these patients, as they experienced significant psychological symptoms going off nicotine. It is unclear whether in the long run these patients achieved abstinence and this in turn improved their outcome. The above study therefore needs to be performed. Until then, the authors would caution pain clinicians against forcing chronic LBP patients to quit smoking unless these patients are willing and ready to do so.

Conclusions

Positive smoking status within chronic LBP patients is associated with poorer treatment outcome (employment) post multidisciplinary pain facility treatment. In addition, chronic LBP smokers' employment status at follow-up is predicted by perceived pain and worker compensation status. Pain clinicians may wish to target smoking chronic LBP patients for additional treatments for pain in order to improve treatment outcome.

References

View Abstract