Demographic, Health, and Behavioral Factors Associated With Smoking in Adults with Type 1 or Type 2 Diabetes Ronald C. Plotnikoff, PhD; Sonia Lippke, PhD; Tricia Prodaniuk, MSc T. Cameron Wild, PhD; Jennifer E. Barrett, MSc Objective: To identify demographic, health, and hehavioral factors associated with smoking hehavior in adults with diahetes. Methods: Canadian adults 18+ years with type 1 (n=697) or type 2 (n=1621) were investigated. Logistic regression analyses were conducted separately for hoth diabetes subgroups. Results: When comparing never versus ever smokers, never versus current smokers, and former smokers who
quit versus current smokers, similarities and differences for demographic, health, and behavioral factors were found for the 2 diabetes subgroups. Conclusions: Diabetes type, demographic, health, and behavioral factors should be considered when tailoring smoking cessation and prevention programs. Key words: smoking behavior, diabetes type, smoking factors Am J Health Behav. 2007;31(l):13-23
C
igarette smoking remains the leading cause of preventable morbidity and mortality.''^ Smoking increases risk of hypertension, ischemic heart disease, chronic obstructive pulmonary disease, stroke, and lung cancer"*'^ and can decrease one's life expectancy by approximately 8 years.'' For people living with diabetes, the health consequences of
smoking are even greater.^'^ Individuals living with this disease have a significantly increased risk of developing a variety of micro- and macrovascular complications, including retinopathy, nephropathy, and cardiovascular disease (CVD).257,8 Several studies report the risk of developing CVD in those with diabetes is twice that of the general population, and smoking may increase this risk.^'^'''"'" MacFarlane reported a 2- to 3-fold increased risk of developing CVD for those Ronald C. Plotnikoff, Professor, Centre for with diabetes who smoke when compared Health Promotion Studies, School of Public to their nonsmoking counterparts.^ FurHealth, University of Alberta, Edmonton, Alberta, ther, reseairch suggests smoking may be Canada. Sonia Lippke, Professor, Psychologie, linked to poor metabolic control,^ limited Freie Universitaet Berlin, Berlin, Germany. Tricia Prodaniuk, Research Coordinator; T. Cameron joint mobility, Cupuyren's contracture, Wild, Associate Professor; Jennifer E. Barrett, and impotence.^'^ Research Assistant, Centre for Health Promotion According to the 2003 Canadian ToStudies, School of Public Health, University of bacco Use Monitoring Survey, approxiAlberta, Edmonton, Alberta, Canada. mately 21% of the general population Address correspondence to Dr Plotnikoff, Cenaged 15 years and older are current smoktre for Health Promotion Studies, School of Public ers.'^ Some research suggests smoking Health, 5-10 University Extension Centre, 8303 prevalence among people living with dia112 Street, University of Alberta, Edmonton, l i ^ ^ Alberta, Canada. E-mail:
[email protected] betes mirrors the general Am J Health Behav.™ 2007;31(l|: 13-23
13
Smoking and Diabetes
Other surveys, however, have reported lower smoking prevalence among people with self-reported diabetes (15.4%).'° The discrepancy in these proportions may be attributed to the fact that smoking is age related. Older individuals are less likely to smoke,® and due to the higher average age amongst diabetes populations, their smoking prevalence can appear lower.'" Notwithstanding these findings, it is nevertheless important to identify factors associated with smoking behavior for people living with diabetes in order to develop effective strategies for smoking cessation in this population. There is some evidence that some demographic and health correlates of smoking status are similar for the general and diabetes populations. As in the general population,'^''^ studies of those with diabetes report that male gender,®''* physical inactivyounger age-groups. ity,^'^ and poor health status' are all associated with smoking behavior. However, little research to date h a s examined whether other factors associated with smoking in the general population, eg, higher body mass index (BMI) and lower income,^""^^ are also associated with smoking among those living with diabetes. Another limitation of extant research is that smoking status and correlates of tobacco use have not been separately examined for type 1 versus type 2 diabetics within the same study. Due to differences in the etiology of type 1 and type 2 diabetes,^' factors associated vwth smoking behavior and subsequent needs for cessation programs may be different in these subpopulations. The purpose of the present study was therefore to identify demographic (age, sex, martial status, education, and income), health (age at diagnosis, CVD occurrence and risk factors, BMI), and behavioral factors (healthy eating and physical activity) associated with smoking in adults with type 1 or type 2 diabetes. METHODS Sample Recruitment Adults (18 years of age and older) were recruited by 2 strategies (Alberta Lorigitudinal Exercise and Diabetes Research Advancement Study: ALEXANDRA).^''^s One thousand nine hundred twenty-eight individuals (609 type 1; 1313 type 2) were recruited from the Canadian Diabetes Association registry. Alberta Chapter.
14
Another 396 individuals (88 type 1; 308 type 2) were recruited from Alberta households using a randomized digit dialing protocol. Individuals from both recruitment strategies completed a self-administered questionnaire on physical activity and other health behaviors over a onemonth period in 2002. The 2 study samples demonstrated no meaningful differences (most P-values >0.05) on demographic, health, and behavioral variables^"'^^ and, consequently, were pooled for subsequent analyses. Therefore, the sample for the current study consisted of 687 type 1 and 1621 type 2 individuals. Further details of the sample recruitment and response rates are reported elsewhere.^''•^^ Measures Demographic factors were measured using questions based on the Statistics Canada 2001 Census^* and included age, sex, ethnic origin, marital status, educational level, and gross annual family income. Health factors were assessed using self-report measures^*'^^'^® and included diabetes type, height and weight (used to calculate BMI), age at diagnosis, CVD (angina, past myocardial infarction), and CVD risk (ie, elevated cholesterol and hypertension). Physical actiinty behavior was assessed using a modified version of the Godin Leisure-Time Exercise Questionnaire (GLTEQ).^' Participants were categorized as active or inactive based upon public health guidelines of achieving the equivalent of at least 150 minutes per week of moderate intensity activj^-y 25.30,31 Healthy eating behavior^'^ was a s -
sessed using one item that asked, "On average, over the past month, how many days per week have you followed your eating plan?" with response options ranging from 0 to 7. Smoking behavior was initially assessed with a standard question "Do you currently smoke cigarettes?" with a yes/no response option. Current smokers were then asked to indicate the number of cigarettes they smoke per day. If participants answered "no" to this question, they were then asked "Have you ever smoked cigarettes?" with a yes/no response option. Responses to these 2 items were used to classify respondents as current smokers, former smokers, or never smokers." Analyses Logistic regressions were performed to
Plotnikoff et al
determine if smoking status was predictable from demographic, health and behavioral factors. Smoking status was defined as never smokers (individuals who had never smoked during their lifetime) and ever smokers (individuals who had smoked during their lifetime). Ever smokers were further differentiated into current smokers (still smoking) or former smokers (quit smoking). Potential differences were examined between (a) never versus ever smokers, (b) never versus current smokers, and (c) former versus current smokers. Logistic regression analyses predicted these different smoking outcomes from demographic (demographic model), health (health factor model), and behavioral variables (behavioral model) separately to determine the unique contributions of variables within each domain. Each of the 3 submodels was adjusted for all variables within each model. Analyses were then completed with all variables (combined model) in order to test the relationship of each variable with smoking behavior controlled for all other variables across the 3 domains. RESULTS Sample Characteristics The sample was primarily Caucasian (type 1: 92.7%; type 2: 89.9%) comprising 53.7% and 48.6% females for the type 1 and type 2 groups respectively. The type 1 group was younger than the type 2 group with respective mean ages of 51.1±17.1 and 62.9±12.1 years. Of the type 1 group, 72.2% were married/partnered, and 76.5% of the type 2 group were married/ partnered. Type 1 individuals had a higher educational level (43% with a university degree vs 34% for the type 2 group), and higher income levels (70% with a gross family income higher than $60,000 vs 56.7% for the type 2 group). The type 1 group had a lower prevalence of heart disease (type 1: 18.2%; type 2: 23.0%), elevated cholesterol (type 1: 36.7% and type 2: 58.7%), and high blood pressure (type 1: 46.8% and type 2: 63.3%). Further, the type 1 group had a lower BMI (type 1: 26.3±4.4; type 2: 29.6±5.9). Type 1 group reported higher levels of physical activity, with 36.3% (versus 28.1% in the type 2 group) of meeting physical activity guidelines. Participants with both type 1 and type 2 reported similar healthy eating behavior. Am J Health Behav.™ 2007;31(l):13-23
Approximately 8% of both the type 1 and type 2 groups were current smokers. Current smokers reported smoking on average 11.9 (SD=7.5) and 13.0 (SD=9) cigarettes per day respectively for the type 1 and type 2 groups. The percentage of former smokers in the 2 groups was 42% and 50% respectively. A total of 46% (type 1) and 38% (type 2) reported never smoking. The sample reflected the Canadian diabetes population for sex, age, BMI, and physical activity status.'" The prevalence of current smokers however was lower in our sample than rates reported for the Canadian population (7.9% versus 15.4% respectively).'" Never versus Ever Smoked Type 1 Diabetes. The demographic model demonstrated that age, sex, and education were significantly associated with being an ever versus a never smoker (Table 1). Specifically, older individuals, females, and individuals with higher educational attainment were more likely to have smoked in their lifetime. The health factor model revealed that ever smokers compared to never smokers were significantly more likely to be older at diagnosis and have a higher prevalence of CVD. Individuals who were diagnosed at an older age had a greater probability to have smoked in their lifetime. Further, people with self-reported CVD were more likely to have smoked. The behavioral model indicated that people who reported higher levels of physical activity were more likely to have never smoked than were persons who were physically inactive. In the combined model (R^ = .13), sex, education, age at diagnosis, and CVD were found to be significantly associated with being an ever versus a never smoker, whereas age and physical activity no longer predicted ever-smoking status. The significant factors revealed that women exhibited 1.79 greater odds of ever smoking compared to men (95% CI: 1.25 - 2.56), and individuals with a university degree were 1.63 times more likely than those without a degree (95% CI: 1.13 - 2.36) to have smoked^ in their lifetime. Persons diagnosed later in life were slightly more likely than those diagnosed younger (OR: 1.02; 95% CI: 1.01 - 1.04), and individuals who experienced CVD were almost twice as likely as those who reported no CVD to have smoked in their lifetime (OR: 1.86;
15
Smoking and Diabetes
Table 1 Logistic Regression Results for Never vs E^ver Smokers Demographic Model
Type 1 Diabetes O=Never Smoker; n=281; l=Smoked in Lifetime; n=335 Odds ratio 95% CI
Variables Age continuous variable (18-92 years) Sex (1 = male, 2 = female) Marital Status (0=no partner; l=partner) Education (1 =no degree; 2=degree) Gross Family Income continuous variable (
