We conducted a cross-sectional survey in which we targeted all adults with SMI who received services from community-based mental health teams in Vancouver, Canada. The vast majority of non-institutionalised persons with a diagnosis of SMI, in this city, are followed by one of these teams (they provide services to almost 6,000 people, more than 1% of Vancouver's population). Each mental health team provides psychiatric assessment and comprehensive treatment through drop-in and outreach services for people in their catchment area. Services include medication management, individual and group therapy, rehabilitation, and education. Many clients receive additional support in the form of rehabilitation programming or housing through contracted agencies.
We sought to obtain a representative sample of people with SMI receiving community mental health services. Because of confidentiality concerns (i.e., disclosure of names and diagnoses without consent), however, we were not permitted to draw a random sample from the population of people receiving services. Consequently, we recruited voluntary participants who were receiving services from seven of the eight mental health teams. Eligible participants were individuals whose health records were flagged as active and who received care from an adult care program. All study participants were living in the community and were able to communicate and be understood in English, Mandarin, Cantonese, or Punjabi.
The research staff visited each community mental health team, provided information about the study, answered questions, and negotiated strategies to access eligible participants. A research assistant recruited participants at the mental health team offices during regular operating hours. The participants were introduced to the survey either through the reception desk personnel or their case managers. The participants could "self refer" to the research staff in response to brochures and flyers available in the office waiting areas. The research staff explained the study in detail, obtained written, fully informed consent, and administered the questionnaire . Upon completion of the questionnaire, the participants received a $10 gift certificate for a local grocery store. Data collection occurred between October 2005 and October 2006, with each mental health team involved for approximately 4-6 months.
Ethical approval was obtained from the Behavioural Research Ethics Board of the University of British Columbia. Approval to conduct the research was obtained from Vancouver Coastal Health, Vancouver Community Health Service Delivery Area.
The questionnaire, which included several scales and items, requiring 20-45 minutes to complete, was administered by the research staff.
The demographic items included: age ("What is your birth date?"), gender ("Do you identify as male, female, trans-gendered or other?"), and ethnic/cultural background ("What would you say is your main ethnic or cultural background?"). The information from this item was used to create a "racialised group" variable ("no" or "yes"). The use of this term is meant to construe the belief that racial classifications are socially constructed and embedded in Eurocentric notions of inferiority, colonization, and prestige . In the study community, people who are Aboriginal, Asian, South Asian or Black tend to be racialised, which has implications for their health . The other demographic variables included: marital status ("What is your current marital status?"), current living situation ("Who do you live with? Alone, with family, friend(s), group home, or other?"), and housing type ("What kind of housing do you live in?" Independent, semi-independent, residential, shelter/hostel, no fixed address, other?), financial support ("In the last month, where have you received money or financial support from? Earned income/paid work, social assistance/welfare, disability benefits, unemployment insurance, pension, savings, alimony/child support, family contribution, panhandling, other"), disposable income ("After paying for housing and food last month, how much money did you have to spend on yourself?"), and income "prioritizing strategies" ("When you have to make decisions about spending money on cigarettes, have you ever chosen to give up anything so that you would have enough tobacco? Have you given up buying food? Coffee? Bus fare? Rent? Medication? Anything else?").
Not all of the participants (15.1%) provided permission to access their medical records. These individuals' diagnostic information was limited to a self-report of the psychiatric diagnosis ("What is your diagnosis?"). For the remainder who provided consent (84.9%), information about their diagnoses was collected from their existing mental health team medical record. Once referred to a community mental health team, all clients are assessed by one of the team's psychiatrists. The psychiatrists typically base their diagnoses on findings of a one-hour assessment interview (that includes mental status examination and case history). DSM IV criteria are used to guide the diagnostic process. A diagnosis is recorded at the time of the client's intake to community mental health services, and then modified as required. For the purposes of this study, the most current diagnosis was recorded.
For the purpose of the analysis, we classified the specific diagnoses as schizophrenia spectrum disorders, mood disorders, or anxiety disorders. A diagnosis of a schizophrenia spectrum disorder included schizophrenia and its subtypes, schizoaffective disorder, delusional disorder, or psychosis not otherwise specified. Mood disorders included diagnoses of bipolar disorder, major depression, manic depression or dysthymia. Anxiety disorders included diagnoses of obsessive compulsive disorder, generalized anxiety disorder, and panic disorder.
Psychiatric symptoms were assessed with the Brief Symptom Inventory (BSI) , which has been validated for use with people living with schizophrenia and is preferred over other scales of psychopathology because it is relatively non-invasive, quick to administer, and suitable for use by research staff . The 18-item scale measures anxiety (e.g., nervousness or shakiness inside), depression (e.g., feeling lonely), and general somatic symptoms (e.g., feeling weak in parts of your body) using a 5-point scale to measure the extent of distress experienced over the past week; the response options were: "not at all," "a little bit," "moderately," "quite a bit," and "extremely." The internal consistency for the Global Severity Index (GSI) has been reported to be strong with a coefficient alpha of .89 . In this study, the scale had a coefficient alpha of .92. We followed the prescribed BSI scoring method: the raw GSI score was calculated by adding the 18 items . If participants had more than 2 item responses missing for any subscale, their scores were not calculated and the case was treated as missing. When participants had 1 or 2 missing items, values were imputed by rounding the mean of the completed items to the nearest whole number. The GSI scores were standardized using T scores with a mean of 50 and an SD of 10 to determine "caseness." Those with GSI scores of 63 or greater were deemed to be at positive risk for psychological distress [14, 16].
Tobacco Use Patterns
Smoking status was determined by asking the participants if they had "ever" smoked, whether they had smoked more than 100 cigarettes in their lifetime, when they smoked their last cigarette, and if they smoked every day . The participants were classified as non-smokers (had never smoked or smoked less than 100 cigarettes), former smokers (had smoked more than 100 cigarettes, but had not smoked in the past 30 days), or current smokers (had smoked more than 100 cigarettes and had smoked in the past 30 days). A binary variable was created with current smoker versus former/never smoker. The participants also were asked, "Do you consider yourself a current smoker?" (The response options were "yes" or "no.") There was excellent agreement between the classification of smoking status based on the number of cigarettes smoked in the past 30 days and the participants' self-reported smoking status (Kappa = .97).
Tobacco use patterns and practices were measured by determining the amount of tobacco smoked each day, the age of smoking initiation  and reasons for tobacco use . Physical health consequences of tobacco use were assessed with the item, "Do you have, or have you had symptoms that you believe were caused or made worse by smoking?" . Items also were included to determine: the primary sources of tobacco procurement ("As you know, cigarettes are expensive and people get them in different ways. Where do you get yours?"), average weekly expenditure on tobacco ("About how much money do you spend on tobacco per week?"), and type of cigarettes smoked ("What kind of cigarettes do you smoke... store bought, roll your own, butts, other?").
Nicotine dependence was measured with the Fagerström Test for Nicotine Dependence (FTND) . This test is appropriate for the assessment of nicotine dependence in smokers with schizophrenia . The coding algorithm yields a total score of 0-10. Scores above 6 are indicative of a high level of dependence. Although widely used, the internal consistency for the FTND scale has been borderline (Cronbach's alpha .67) ; in this study, the Cronbach's alpha was .50. In addition to using this scale, the participants were asked to rate their tobacco addiction using a self-rated addiction scale of 0-10, where 0 was "not at all" addicted and 10 was "extremely" addicted. They also were asked about using tobacco to manage their psychiatric symptoms: "Some people use smoking to cope with their symptoms, such as having anxiety or hearing voices. How often do you smoke to cope with symptoms?" The item was scored with a 4-point scale rated as "not at all," "a little," "somewhat," or "a great deal." Another open-ended question asked, "What symptoms do cigarettes help you manage?"
Comorbid substance use was assessed with items from the substance use section of the Addiction Severity Index (ASI), originally developed for clinical purposes , . The ASI has seven sections measuring various aspects of an individual's life that may be affected by substance use. For research purposes, the use of individual items from the substance use section of the ASI has been found to be reliable, valid, and valuable . The participants were asked, "How many days in the past month (last 30 days) did you use...any alcohol? Alcohol to get drunk? Heroin (smack, junk)? Methadone? Opium, codeine, or pain killers like Tylenol 3? Sedatives, hypnotics or tranquilizers like Valium or Xanax? Cocaine or crack? Amphetamines, like speed, E or meth? Marijuana (weed, pot)? Hallucinogens, like LSD or mushrooms? Inhalants, like glue, paint thinner or gas? Any other substances? Specify." The ASI results were reported as number of days and were categorized into "no, none" or "yes, 1 or more days" because of the participants' infrequent regular use and the distributional properties of their responses .
A total of 788 people participated in the study, which represents approximately 20% of the clients who received care from the 7 community mental health teams. The data from these clients were cleaned and screened before analysis to ensure missing data were random in occurrence and that all data were within their excepted ranges. Responses from 59 (7.5%) individuals were excluded because they did not have a clear psychiatric diagnosis. Descriptive analysis of the sample (N = 729) employed chi square tests to determine the associations between psychiatric diagnosis and the categorical study variables. Independent sample t-tests employing Levine's test for equality of variance were employed to examine the relationships between psychiatric diagnosis and the continuous variables. We employed Hosmer and Lemeshow's model-building process to determine the variables that were associated with current smoking status (current smoker vs. former/never smoker) . First, we employed univariate logistic regression analyses to identify the study variables associated with smoking status and conducted these analyses for the entire sample and for men and women, separately. In the second step, variables that were associated with smoking status at p ≤ .25 were included in the multivariate logistic regression models (all participants and gender-specific). To obtain the most parsimonious and stable models, we then trimmed them by removing statistically non-significant variables sequentially by examining the Wald statistic and comparison of the likelihood ratios. If the likelihood ratio test was significant when a non-significant variable was removed (i.e., p < .05), then the variable was added back to the model. Once the main effects models were finalized, all possible interactions between diagnostic category and the other variables were examined. All analyses were conducted with IBM SPSS Statistics 18.