|Home | About | Journals | Submit | Contact Us | Français|
The majority of cigarette smokers have a lifetime diagnosis of substance abuse and/or mental illness, and treatment outcomes for smokers with these comorbidities are generally reported to be worse than for smokers without co-morbidities. We sought to examine the effect of specific substance abuse/mental illness diagnoses compared to one another on treatment outcomes.
A retrospective chart review of naturalistic treatment for Tobacco Dependence was performed on male smokers (n= 231) who enrolled in the Greater Los Angeles Veterans Affairs Mental Health Clinic Smoking Cessation Program over a 1.5 year period. Subjects in this program underwent comprehensive treatment for Tobacco Dependence (including, but not limited to, group psychotherapy, nicotine replacement therapy, and bupropion HCl). Quitting smoking was defined as a report of at least 1 week of abstinence and an exhaled carbon monoxide less than 8 parts per million at the final clinic visit.
Of the total group, 36.4% (84/231) quit smoking at the end of treatment. Quit rates were affected by the presence of specific diagnoses, with smokers with a history of Alcohol Abuse/Dependence or Schizophrenia/Schizoaffective Disorder having poorer response rates than smokers without such diagnoses. Other substance abuse and mental illness diagnoses did not affect quit rates.
Lower quit rates among patients with Alcohol Abuse/Dependence or Schizophrenia/Schizoaffective Disorder may be due to the severity of these conditions, and suggests that specialized treatment is needed for these populations of smokers. Smokers with most co-morbid diagnoses are successfully treated with standard treatment methods.
Cigarette smoking is common among people with a lifetime history of substance abuse/dependence and/or mental illness as assessed in both clinical1 and population-based2,3 samples. For substance abuse/dependence, tobacco use rates are reported to be as high as 90% among active alcoholics,4 80% among cocaine abusers,5 92% among methamphetamine abusers,6 and 82% to 94% among opiate addicts6-8 (compared to the general United States population rate of 23%9). In addition, marijuana smokers are five times more likely than non-marijuana users to smoke tobacco cigarettes.10 Cigarette smoking contributes greatly to morbidity and mortality among patients with alcohol11 and drug12 dependencies, making it vital to understand better the complex relationship between alcohol/drug dependence, mental illness, and response to smoking cessation treatments.
In general, substance abusers are less likely than non-substance abusers to quit smoking with standardized treatment.13 Specifically, both nicotine gum14 and patch15 have been found to be less effective in alcoholic than non-alcoholic smokers, especially for long-term outcomes. Although worse outcomes for substance abusers are fairly consistently reported, there are three studies utilizing the nicotine patch for smoking cessation that demonstrate similar outcomes for patients with and without alcohol abuse/dependence histories. Two studies showed similar outcomes for alcoholics in treatment compared to non-alcoholic controls,16,17 while a high rate of smoking abstinence (51%) was achieved in alcohol-dependent smokers in sustained full remission with a specialized tailored patch treatment program.18 Patients with opiate-dependence histories have been reported to have been treated with nicotine patch in combination with cognitive behavioral therapy or contingency management19,20 with moderate success. We are aware of no prior studies that specifically examined outcomes of smoking cessation treatment (including nicotine replacement therapy, bupropion HCl, and group psychotherapy) among cocaine-, methamphetamine- , or marijuana- dependent cigarette smokers.
Cigarette smoking is also common among patients with mental illnesses other than substance abuse/dependence. Prevalence rates for tobacco use in clinical samples in schizophrenics approach 90%,21-23 43% to 82% in bipolar patients,2,3,22,24,25 and 45%-60%2,26 in patients with PTSD. Patients with a history of depression are at increased risk for nicotine dependence (about 50%)2,3,27 and it is reported that a history of regular smoking occurs more frequently among individuals who have experienced a major depressive episode at some time in their lives.28
Treatment courses for smokers with mental illness have proved challenging.29 Patients with severe mental illnesses been shown to have low rates of smoking cessation in a large naturalistic study.30 Schizophrenic smokers represent one subset of mentally ill smokers that have been particularly difficult to treat,31 though treatment with bupropion HCl has been shown to improve smoking abstinence rates in schizophrenic smokers.32-34 Patients with Major Depressive Disorder have also shown lower success rates with treatment,35 though bupropion HCl has been found to be efficacious in smokers with this illness as well.36 Bupropion has been demonstrated to be more effective than placebo in treatment of veterans with both post-traumatic stress disorder (PTSD) and nicotine dependence.37 We are aware of no studies that have examined efficacy of smoking cessation treatment among bipolar patients.
While standardized treatments (as described above) have been well-studied in the substance abusing population, less is known about treatment for nicotine dependence in a naturalistic setting, where first-line (such as nicotine replacement therapy, bupropion HCl, and group psychotherapy) and second-line (such as tricyclic antidepressants and clonidine) treatments are used alone or in combination. While it is known that substance abuse and mental illness generally predict worse outcomes in treatment, a comparison of treatment outcomes between specific substance abuse and mental illness conditions has not yet been reported. Based on the scientific literature and knowledge of the individual conditions, we hypothesized that patients with more severe drug dependencies (e.g., polysubstance dependencies) and more pervasive mental illnesses (e.g., schizophrenia-spectrum disorders) would have more difficulty quitting smoking during naturalistic treatment than those without these conditions. Our aim in conducting this research was to understand better the specific substance abuse/dependence and mental illness diagnoses, and their interactions, which affect quit rates among cigarette smokers in a naturalistic setting.
A retrospective chart review of naturalistic treatment for Tobacco Dependence was performed on a convenience sample of 231 male cigarette smokers who enrolled in the Greater Los Angeles Veterans Affairs Mental Health Clinic Smoking Cessation Program. This smoking cessation program enrolls only smokers with a current or a history of substance abuse/dependence and/or mental illness.
All patients in this chart review were diagnosed with Nicotine Dependence during their participation in the smoking cessation program, based on a semi-structured interview using DSM-IV criteria38 and a score of 3 or greater on the Fagerström Test for Nicotine Dependence (FTND).39,40 All subjects were treated with practical group counseling (45 minutes per week) based on the relapse prevention model41 for a standard 6- to 12- week course, in combination with nicotine replacement therapy, bupropion HCl, or both (based on patient preference in consultation with a clinic physician [R.B.G. or A.L.B.]). In the practical group counseling, subjects received education about tobacco and the treatment of Nicotine Dependence, monitoring of exhaled carbon monoxide levels, practical advice on smoking cessation techiniques, problem-solving exercises, social support, and co-mentoring. Medication treatments were individualized based on patient preference, and subjects received the nicotine patch alone, bupropion HCl alone, a combination of the nicotine patch and bupropion or no pharmacologic treatment. Exhaled carbon monoxide (CO) measurements were monitored on all patients on a weekly basis with the commonly used Bedfont EC-50 Microsmokerlyzer II device, as a rough measure of recent cigarette usage.
Subjects were considered to have quit smoking at the end of program participation, if they reported at least 1 week of smoking abstinence and had a final exhaled CO level of ≤ 8 parts per million. These criteria are similar or identical to other treatment studies of tobacco dependence.42,43
The following information was collected from the initial treatment visit and from a review of the patient's history prior to coming to treatment: age, gender, number of cigarettes smoked, number of years smoking, number of quit attempts, longest quit period, Fagerström Test for Nicotine Dependence (FTND)39,40 and Beck Depression Inventory (BDI)44 scores, previous mental illness and substance abuse/dependence diagnoses, length of substance abuse/dependence, length of substance abstinence, medications used at the onset of treatment, and co-morbid medical conditions. Information collected from the treatment itself included: number of clinic visits, medication received (nicotine replacement therapy usually with the nicotine patch versus bupropion HCl versus both versus neither), exhaled CO levels throughout treatment, and number of cigarettes reported per day for the last week of treatment.
In order to explore the central question of this study (namely, which mental illness or substance abuse/dependence diagnoses predict treatment outcome in a naturalistic setting), we performed logistic regression (SPSS version 13.0, Chicago, IL) using the fewest possible variables based on what is known about prediction of treatment response in smokers. Quit status at the end of treatment was the dependent variable. Independent variables were four mental illness diagnoses (Major Depressive Disorder, Bipolar Disorder, Post-Traumatic Stress Disorder, and Schizophrenia/Schizoaffective Disorder) and five substance abuse/dependence diagnoses (alcohol, marijuana, cocaine, methamphetamine, and opiates). Age and number of cigarettes smoked per day were used as confounding independent variables in all logistic regression model because these variables are known to affect treatment outcome.45 Diagnoses that were seen in fewer than 5% of the subjects (e.g., Panic Disorder, Generalized Anxiety Disorder, and hallucinogen abuse) were excluded from the analysis. The Bipolar Disorder group included patients diagnosed with either Bipolar I or II disorder. Demographic and treatment variables are presented as the mean ± standard deviation. We first evaluated each predictor in a separate logistic regression analysis, controlling for age and cigarettes smoked per day. Then, we evaluated all the diagnostic variables in a single multiple logistic regression model. We report the significance tests from both the separate and multiple models and odds ratios. In all cases, the chi-square tests are for the increment over and above the covariates. The Homer-Lemeshow test was used for global goodness of fit testing.
The 231 subjects were middle-aged (49.1 ± 8.3 years old), smoked moderately at the time of treatment initiation (17.6 ± 9.9 cigarettes per day), had lengthy smoking histories (30.9 ± 9.9 years), few quit attempts (2.4 ± 2.6), moderately long prior periods of smoking abstinence (1.7 ± 3.6 years), moderate tobacco dependence (FTND scores 5.1 ± 2.2), and mild depressive symptoms (BDI scores 5.1 ± 3.8). For the substance abuse diagnoses studied here, more than half of the subjects had been diagnosed with Alcohol Abuse/Dependence (n = 184, 80%) or Cocaine Abuse/Dependence (n = 133, 58%), while fewer were diagnosed with Marijuana (n = 78, 34%), Methamphetamine (n = 50, 22%), and Opiate Abuse/Dependence (n = 47, 20%). For the mental illness diagnoses, more than half of the sample had been diagnosed with Major Depression (n = 131, 57%), while smaller numbers were diagnosed with Bipolar Disorder (n = 28, 12%), Post-Traumatic Stress Disorder (n = 50, 22%), and Schizophrenia/Schizoaffective Disorder (n = 39, 17%). The most common patterns of co-occurring mental illness and addiction were Alcohol Abuse/Dependence, Cocaine Abuse/Dependence and Major Depression (n= 17, 7.4%), Alcohol Abuse/Dependence, Cocaine Abuse/Dependence, Marijuana Abuse/Dependence and Major Depression (n=13, 5.6%), and Alcohol Abuse/Dependence and Cocaine Abuse/Dependence (n=11, 4.8%).
84 of 231 participants (36.4%) met criteria for having quit smoking at the end of treatment. Table 1 summarizes the separate logistic regression analyses, and Table 2 summarizes the multiple regression analysis. The overall multiple regression model yielded a likelihood-ratio chi-square = 16.83, df = 9, p = .05 for the increment of the nine predictors over and above the 2 covariates. Two of the substance abuse diagnoses were significant in the multiple regression analysis: Alcohol Abuse/Dependence (Wald Chi-Square = 4.03, df = 1, p= .05, OR=2.06) (associated with a worse quit rate), and Marijuana Abuse/Dependence (associated with a better quit rate) (Wald Chi-Square = 4.07, df = 1, p = .04, OR = 0.51) (Figure 1). The diagnoses of Schizophrenia /Schizoaffective Disorder were significant in both the separate logistic regression (chi-square = 5.74, p=0.02, OR=2.84) and multiple logistic regression models (chi-square = 7.40, p<0.01 , OR = 3.80, both df = 1) (Figure 2). Individuals in this diagnostic category were less likely to quit successfully. The Homer-Lemeshow goodness-of-fit test indicated that the model prediction did not significantly differ from the observed valuses (p=.37) The total number of substance abuse/dependence diagnoses did not have a linear relationship with the chance of quitting smoking (chi-square = 0.32, df = 1, p = .57), indicating that specific diagnoses rather than more diagnoses, were associated with treatment responsiveness.
The total number of substance abuse/dependence diagnoses did not have a linear relationship with the chance of quitting smoking (chi-square = 0.32, df = 1, p = .57), indicating that specific diagnoses rather than more diagnoses, were associated with treatment responsiveness.
The pharmacologic treatment used in the study varied with patient preference. The majority of the subjects were treated with the nicotine patch alone (n = 140), while many used bupropion HCl alone (n = 53) and fewer used the combination of nicotine patch plus bupropion HCl (n = 30). Treatment with both nicotine patch and bupropion HCl together resulted in a quit rate of 33%, which was not higher than the quit rates of subjects treated with monotherapy.
Results of this retrospective chart review demonstrate that the overall smoking cessation rate for patients with mental illness and/or substance abuse/dependence treated in a naturalistic setting are similar to those of published reports for controlled treatment in more general populations of smokers.15,20,46,47 For specific diagnoses associated with treatment responsiveness, our data demonstrate that the absence of an alcohol dependence history predicts better response to treatment, while the presence of schizophrenia/schizoaffective disorder predicts poorer outcome to treatment. The presence of marijuana abuse predicts better response to treatment.
A history of alcohol dependence might lead to lower quit rates than the absence of this history for several reasons. There is considerable evidence that chronic alcohol use leads to structural, physiological and functional brain changes,48-50 along with neuropsychological impairment.51 Chronic alcohol use leads to atrophy of the frontal lobes,48 hypofrontal brain metabolism52 and electrophysiological abnormalities.53 Such frontal lobe damage is associated with impulsivity, impaired planning, poor problem solving, and impaired insight and judgment.54 This type of brain damage (even in a subtle form) could account for the lower quit rates in smokers with a history of alcohol dependence, since behavioral modification (part of the treatment program here) might be more challenging in an impulsive subject with impaired coping and planning skills.
Cognitive deficits associated with alcohol dependence are well documented. It has been shown that between one-half and two-thirds of newly abstinent alcoholics exhibit cognitive impairment in the first few months of sobriety.51 These deficits may persist for years or indefinitely after detoxification in some patients, and include impairment in new learning, executive functioning, visual spatial abilities, and perceptual-motor integration.55 Specifically, verbal learning56,57 and retention,58 as well as abstract reasoning59,59-61 have been shown to be impaired in patients with a history of alcoholism. These specific deficits may lead to suboptimal participation in treatment, where verbally acquired directions for treatment and abstract reasoning are important for such components of treatment as medication instructions, suggestions for behavioral modification from group therapy, training in coping with stress, and in the identification of triggers.
Our data also support prior work31 which points to Schizophrenia and Schizoaffective disorders as illnesses that are particularly resistant to smoking cessation treatment. A central feature of these illnesses may be negative symptoms, such as avolition, alogia, and blunted affect.38 Also, similar to alcohol abuse/dependence, structural, physiological and functional deficits are seen in the schizophrenic brain, such as lower frontal lobe volume62 and frontal electrophysiological abnormalities.63 As noted above, structural and functional impairments of the frontal lobes are associated with impulsivity, poor planning and limited problem solving. Additionally, schizophrenic subjects have been identified as having moderately impaired executive functioning and more severely impaired verbal learning and memory when compared with controls.64 Consistent with the functional consequences for such cognitive dysfunction, one study demonstrated that the presence of prefrontal executive function deficits in spatial workingf memory and Wisconsin Card Sorting Test performance, prior to a quit attempt, was associated with smoking cessation treatment failure in smokers with schizophrenia, but not controls.65 They also have low Global Assessment of Functioning scores,66 a measure that includes social functioning, and a higher level of social support has been shown to be a positive predictor for smoking cessation.67 Taken together, these features (negative symptoms, frontal lobe dysfunction, and low global functioning scores) all point to potential mediating factors in the poor response people with Schizophrenia/Schizoaffective Disorder show to comprehensive treatment.
In this study, a history of Marijuana Abuse/Dependence was associated with slightly higher tobacco abstinence rates, when compared to other substance abuse and mental illness diagnoses. Many of the subjects were currently in treatment or had a history of treatment for substance abuse/dependence. Having a history of successful abstinence from a drug of abuse that was experienced through smoking (rather than intravenous or intranasal use) may have conferred a slight advantage in this group of tobacco users. Since they had been able to discontinue smoking marijuana in the past, perhaps this led to successful discontinuance of tobacco use due to higher confidence in their own abilities to overcome a similar habit.
One limitation of this study stems from the sample of male veteran subjects. It is known that the rates of substance abuse68,69 and affective disorders68 are higher in the veteran population when compared with similar gender and age-matched patients in the private sector. Furthermore, male veterans with severe mental illnesses, such as schizophrenia, tend to be older and to have had more inpatient hospitalizations,70 which suggests that the sample used in this study may have been older and more ill than patients who might be found in other types of healthcare systems. A second limitation of the study is that long-term follow up data (6 months to 1 year) was not available. A third limitation of the study is its retrospective design. The retrospective set-up of the study was the most reasonable and practical design for a naturalistic investigation of the data. While we are aware that such study designs may introduce incomplete study groups and interpretive bias during the chart review, these limitations were minimized by remaining cognizant of them.71,72 Another limitation of this study is the same as that of most naturalistic treatment studies, namely that subjects were not randomly assigned to treatments and clinicians were not blinded to the treatments administered. This lack of random assignment may have led to subjects being assigned to particular treatments which may have altered their possibility of quitting (e.g., it is likely that patients with Bipolar Disorder were less likely to receive bupropion HCl because mania is a potential side effect of this antidepressant). However, our data reflect treatment in a naturalistic setting, which may prove informative for real-life treatment scenarios.
In summary, this study identified several predictors of the success of smoking cessation in this naturalistic setting. While prior work has been done to start to tailor treatments to schizophrenic smokers,73 our data indicate that further research is needed to optimize treatments for smokers with psychotic illnesses and/or alcohol dependence. Both schizophrenic and alcohol dependent smokers may be better served with modified treatments15 that are specialized to take into account the needs of these specific populations.
Presented in part at the Society for Research on Nicotine and Tobacco Annual Meeting, Orlando, Florida, February 17, 2006. Supported by the National Institute on Drug Abuse (A.L.B. [R01 DA15059 and DA20872]), a Veterans Affairs Type I Merit Review Award (A.L.B.), the National Alliance for Research on Schizophrenia and Depression (A.L.B.), and the Tobacco-Related Disease Research Program (A.L.B. [11RT-0024]).
Risa B. Gershon Grand, UCLA Department of Psychiatry & Biobehavioral Sciences and Greater Los Angeles Veterans Affairs Healthcare System, Los Angeles, California, USA.
Sun Hwang, UCLA Department of Psychiatry & Biobehavioral Sciences, Los Angeles, California, USA.
Juliette Han, Greater Los Angeles Veterans Affairs Healthcare System, Los Angeles, California, USA.
Tony George, Yale University School of Medicine, New Haven, Connecticut, USA.
Arthur L. Brody, UCLA Department of Psychiatry & Biobehavioral Sciences and Greater Los Angeles Veterans Affairs Healthcare System, Los Angeles, California, USA.