Smoking habits and incidence of cardiovascular diseases in men and women: findings of a 12 year follow up among an urban Eastern-Mediterranean population

The TLGS is an ongoing community-based study initiated in 1999, and designed to be continued for at least 20 years. The main goal of study is to assess risk factors and different aspects of life style in relation with NCDs among a representative urban Tehranian population [22]. Details of its goals and design have been published previously [22, 23].

The current analysis has been conducted on all TLGS participants who participated in the study between 1999 and 2002; those aged under 20 years, those with positive CVD history, those with missing smoking data and those lost to follow up were excluded. A final sample of 10,400 adult individuals (4378 men and 6022 women) was followed for a median of 12 years. The sampling frame of the current analysis is shown in Fig. 1. The study was approved by the research ethics committee of the Research Institute for Endocrine Sciences (RIES), Shahid Beheshti University of Medical Sciences. Written informed consents were obtained from all participants.

Data on socio-demographic information (age, education level, marital status and occupation), smoking habits and exposure to ETS and any familial history of NCDs were collected by trained interviewers. For physical activity, data was collected using a validated, reliable questionnaire entitled the Modifiable Activity Questionnaire (MAQ) [24]. Psychometric properties of the Iranian version of MAQ have been assessed in the TLGS population and showed acceptable validity and high reliability [25]. Weight was measured by digital scales and was recorded to the nearest 100 g, without shoes and with light clothing. Height was measured standing, without shoes. Body mass index (BMI) was calculated by dividing weight (kg) by square of height (m²). After participants rested for 15 min, a physician measured blood pressure using a mercury sphygmomanometer with appropriate cuff size for each individual. Systolic and diastolic blood pressure were measured twice, at least 30-s apart and the mean of the two measurements was recorded as the participant’s blood pressure. After 12 to 14 h of overnight fasting, blood samples were collected between 7:00 and 9:00 AM to be sent to TLGS research laboratory and biochemically analyzed on the same day. Levels of fasting blood glucose (FBS), total cholesterol, high density lipoprotein (HDL-c) and triglycerides (TG) were enzymatically measured. Low density lipoprotein (LDL-c) was calculated using Friedewald formula [26]. The detailed process of socio-demographic, anthropometric and biochemical assessments has previously been published elsewhere [22, 23].

Male participants were categorized into the six following groups: 1) never smokers, 2) passive smokers (exposure to ETS), 3) ex-smokers (quitter since at least a month before interview), 4) passive and ex-smokers (quitters exposed to ETS), 5) active occasional smokers (non-routine smokers) and 6) active daily smokers. In another grouping pattern, based on number of cigarettes smoked per day, active male smokers (whether occasional or daily) have been further subdivided in to three groups: ≤10, 11–20 and ≥ 21 cigarettes per day. Ex-smokers were also subcategorized into those who quit smoking in less than 15 years, and those that had quit 15 years ago or earlier, regardless of passive smoking. Female participants were categorized as never smokers, passive smokers (same as males) and ever smokers (active and ex-smokers pooled together). Different categorization in men and women in the current study was mainly due to the limited number of women who reported themselves as active (n = 146) or past smokers (n = 94) and consequently the limited number of CVD events that occurred in these groups (n = 15) through the study period.

Participants were followed up annually for any cardiovascular event through phone call interviews by trained nurses. A physician confirmed the diagnosis by home visits or data collection from the medical file or death certification, in case of mortality. A committee consisting of an internist, endocrinologist, cardiologist, epidemiologist and other professionals when needed, confirmed the outcome data; CHD event was defined as any definite myocardial infarction (MI) (diagnostic electrocardiogram (ECG) and biomarkers), probable MI (positive ECG findings and cardiac symptoms plus missing biomarkers or positive ECG findings plus equivocal biomarkers), sudden cardiac death, unstable angina pectoris (new cardiac symptoms or changing symptom pattern and positive ECG findings with normal biomarkers) and angiographically proved coronary artery disease [27]. The definition of CVD event was that of CHD, plus fatal or non-fatal stroke (defined as a new neurological deficit lasting ≥24 h). Details of outcome data analysis and definitions have been described elsewhere [22, 23, 28].

All analyses and baseline characteristics are classified by sex. Continuous variables are expressed as mean ± (SD) for normally distributed and as median (Q1-Q3) for skewed ones. To compare continuous variables between different CVD groups, Student t-test and Mann-Whitney test were used for normally distributed and skewed variables, respectively. In different smoking groups, continuous variables were compared by ANOVA in normally distributed- and by Kruskal-Wallis test in skewed ones. Categorical variables are expressed as number (%) and have been compared with Chi-square test, both within different smoking groups and CVD groups. Cox proportional hazard model was used to study the association between different smoking patterns and cardiovascular outcomes. The follow-up duration was defined as the period starting from entrance to study and ending in occurrence of any predefined CVD/CHD event or any attributed mortality or censoring, whichever happened first. Censoring was defined as loss to follow-up or mortality, not attributable to CVD or CHD. Univariate regression model was first performed for all potential confounding factors and those with p-values less than 0.2 were selected for multivariate regression. Adjustment of potential confounding factors was conducted in three cumulative models; the first model was adjusted for age, the second for age and cardio-metabolic risk factors (BMI, systolic and diastolic blood pressure, fasting blood sugar, total cholesterol and triglycerides) and the third model was adjusted for all the above mentioned variables, plus socio-economic factors (education level, reference: college degree; marital status, reference: married and occupation, reference: employed). Within each model, adjusted hazard ratios (HRs) of outcome events were estimated in men and women with 95% confidence intervals (95% CIs) for each smoking pattern, given never smokers as reference. Proportionality assumption of Cox models was assessed using the Schoenfeld residual test and confirmed. Statistical analyses were performed by STATA software version 12, and SPSS software version 15. P-values below 0.05 were considered statistically significant.