Interaction between smoking and diabetes in relation to subsequent risk of cardiovascular events

Background Whether smoking modifies the associations of diabetes and risk factor management with subsequent risk of cardiovascular disease (CVD), and whether the smoking related CVD risk differs among people with and without diabetes are unclear. This study aimed to examine the associations and interactions of smoking, diabetes, and risk factor management in relation to incident CVD. Methods This nationwide, population-based, prospective cohort study of 20 communities from various geographic regions recruited adults aged 40 years or older during 2011–2012. The follow-up survey was conducted between 2014 and 2016. This study included 126,181 participants who were free from CVD at baseline. Results Study participants included 19,397 current smokers (15.4%), 6,049 former smokers (4.8%), and 100,735 never smokers (79.8%). Mean (SD) age ranged from 55.8 (8.6) years to 60.7 (9.1) years. Compared with never smokers, heavy smokers exhibited a greater risk of CVD events among participants with diabetes (multivariable-adjusted hazard ratio [HR], 1.45; 95% CI, 1.17–1.78) than among participants without diabetes (HR, 1.20; 95% CI, 1.01–1.42; P for interaction = 0.006). Compared with participants without diabetes, participants with diabetes who were never smokers and had 5 or more controlled risk factors showed no significantly excess CVD risk (HR, 0.93; 95% CI, 0.71–1.22), but the cardiovascular benefits from risk factor management were counteracted among participants with diabetes who were current smokers (HR, 1.28; 95% CI, 0.77–2.14) or former smokers (HR, 1.22; 95% CI, 0.66–2.28). Conclusions Smoking and diabetes interacted with each other in relation to increased risk of CVD events, and the beneficial effect of risk factor management on CVD risk among participants with diabetes was attenuated by current or former smoking. Supplementary Information The online version contains supplementary material available at 10.1186/s12933-022-01447-2.


Introduction
Cigarette smoking is the leading preventable cause of morbidity and mortality worldwide, responsible for 11.5% (equivalent to 6.4 million) of global deaths annually [1,2]. In China, smoking has been ranked as the most important lifestyle risk factor contributing to deaths and disability-adjusted life-years [3]. Smoking substantially increases the risk of cardiovascular disease (CVD) in both the general population and patients with type 2 diabetes [4][5][6]. Compared with people without diabetes, patients with diabetes have a twofold additional hazard of CVD, occurring earlier and with greater severity [7][8][9]; meanwhile CVD remains the most common cause of mortality among patients with diabetes [10]. Therefore, interventions to prevent or quit smoking, together with cardiovascular risk factor management such as glycemic control and the management of blood pressure and cholesterol, are key recommendations for the prevention of CVD in diabetes [11,12].
Smoking prevention and cessation interventions compare favorably with the management of other cardiovascular risk factors, in terms of both health improvements and cost savings [13]. However, a considerable proportion of people are likely to continue smoking even after a diagnosis of diabetes [14], and the successful rate of smoking cessation interventions for smokers with diabetes is not ideal [15]. Evaluating the interplay between cigarette smoking and diabetes status in relation to CVD risk is of the essence for prioritizing precise clinical and public health strategies to reduce the burden of CVD among patients with diabetes. Thus far, whether current smoking or smoking cessation modifies the associations of diabetes and risk factor management with the incidence of CVD, and whether the smoking-related CVD risk differs among people with and without diabetes are still unknown.
To fill this knowledge gap, we investigated the bidirectional interactions of active smoking and smoking cessation with diabetes status on subsequent risk of CVD events, and examined the association of cardiovascular risk factor management in diabetes with the risk of CVD across smoking status categories in a nationwide prospective cohort study.

Study participants
The China Cardiometabolic Disease and Cancer Cohort (4C) Study is a multicenter, population-based, prospective cohort study [16,17]. The baseline survey was conducted between 2011 and 2012, and 19,3846 men and women aged 40 years or older were recruited from local resident registration systems of 20 communities from various geographic regions in China to represent the general population. The follow-up survey was conducted between 2014 and 2016; all participants were invited to attend an in-person visit, and 170,240 participants (87.8%) were followed up. The analysis included 126,181 participants who had complete baseline data on cigarette smoking, diabetes status, and covariates; were free from CVD at baseline; and had complete ascertainment of CVD events at the follow-up visit. This study was approved by the Medical Ethics Committee of Ruijin Hospital, Shanghai Jiao Tong University. All study participants provided written informed consent.

Data collection
Data collection was performed in local community clinics by trained study personnel according to standard protocols. Questionnaires comprising information on demographic characteristics, dietary and lifestyle factors, family history, and medical history were administered by trained interviewers. Education attainment was classified as less than high school (< 9 years) and high school or further education (≥ 9 years). A food frequency questionnaire was used to collect habitual dietary intake by asking the consumption frequency and portion size of typical food items during the previous 12 months. International Physical Activity Questionnaire was used to assess physical activity [18], and Metabolic equivalent (MET) was calculated to evaluate average weekly energy expenditure [19]. Physical activity was classified as active (≥ 600 METmin/week), insufficiently active (> 0 to < 600 MET-min/ week), and inactive (0 MET-min/week). Average alcohol consumption was calculated by multiplying amount of alcohol consumed per drinking day by frequency.
Height and body weight were measured according to a standard protocol, and body mass index (BMI) was calculated as body weight in kilograms divided by the square of height in meters.
Three measurements of systolic blood pressure (SBP) and diastolic blood pressure (DBP) were obtained by using an automated electronic device (OMRON Model HEM-752 FUZZY, Dalian, China) in a seated position after at least a 5-min quiet rest, and the 3 readings were averaged for analysis. Hypertension was defined as SBP ≥ 140 mmHg, or DBP ≥ 90 mmHg, or taking antihypertensive medication.
All participants underwent a 2-h, 75-g oral glucose tolerance test (OGTT) after an overnight fast of at least 10 h, and blood samples were collected at 0 and 2 h during the test. Fasting and 2-h plasma glucose concentrations were measured locally using a glucose oxidase or hexokinase method within 2 h after blood sample collection under a stringent quality control program. Finger capillary whole blood samples were collected by the Hemoglobin Capillary Collection System (Bio-Rad Laboratories, Hercules, CA, USA) and were shipped and stored at 2℃ to 8℃ until glycated hemoglobin A1c (HbA1c) was measured within 4 weeks after collection by high-performance liquid chromatography using the VARIANT II Hemoglobin Testing System (Bio-Rad Laboratories) at the central laboratory in the Shanghai Institute of Endocrine and Metabolic Diseases, which was certificated by the National Glycohemoglobin Standardization Program and the College of American Pathologists Laboratory Accreditation Program.

Assessment of cigarette smoking, diabetes, and risk factor management
Information on cigarette smoking was collected by a standard questionnaire. Current smokers were defined as participants who reported smoking cigarettes every day or almost every day, with at least 7 cigarettes per week for at least 6 months [2]. Intensity of current smoking was classified into light (< 10 cigarettes per day), moderate (10-< 20 cigarettes per day), and heavy (≥ 20 cigarettes per day) levels by number of cigarettes smoked [22]. Former smokers were defined as participants who reported not currently smoke cigarettes but had smoked at least 7 cigarettes per week for at least 6 months in a lifetime [2]. Former smokers were classified into short-term quitters (quit for < 4 years), medium-term quitters (quit for 4-< 8 years), and long-term quitters (quit for ≥ 8 years) by years since smoking cessation. Never smokers were defined as participants who reported never smoked or had not smoked cigarettes regularly (< 100 cigarettes in a lifetime) [2]. Passive smoking was defined as exposure to 1 or more smokers during childhood or adulthood.

Ascertainment of cardiovascular events
The outcome was the composite of incident fatal or nonfatal CVD events, including cardiovascular death, myocardial infarction, stroke, and hospitalized or treated heart failure. The detailed definitions of CVD events have been described previously [16]. Deaths and clinical outcomes were collected from local vital registries of the National Disease Surveillance Point System and National Health Insurance System. Two members of the outcome adjudication committee independently verified each clinical event and assigned potential causes of death, and discrepancies were adjudicated by discussions involving other members of the committee. The members of the outcome adjudication committee were unaware of the baseline clinical characteristics of each participant.

Statistical analysis
Baseline characteristics of participants according to smoking status were summarized as means with standard deviations (SDs) for continuous variables and numbers with percentages for categorical variables. In the timeto-event analysis, participants were censored at the date of CVD diagnosis, death, or the end of follow-up, whichever occurred first. Person-time was calculated from the enrollment date to the censoring date for each participant. All participants had complete data of cigarette smoking, diabetes status, and covariates; missing data for risk factor variables were deleted from the analyses.
Cox proportional hazards models were used to calculate hazard ratios (HRs) and 95% confidence intervals (CIs) for CVD events associated with current smoking and smoking cessation among participants with and without diabetes, with adjustment for traditional CVD risk factors as well as baseline BMI, change in BMI during the follow-up period, and passive smoking. Associations of diabetes as well as elevated fasting glucose, elevated OGTT-2 h glucose, and elevated HbA1c with CVD events among current smokers, former smokers, and never smokers were also established. Multiplicative interactions between smoking and diabetes on CVD events were tested by including the product term (e.g., smoking status category × diabetes category) as well as smoking status and diabetes status in the models. We performed additional analyses to assess the associations and interactions between smoking and diabetes on nonfatal CVD events and CVD mortality. The numbers of well-managed risk factors were summed for each participant to represent an overall management of all 6 risk factors. Associations of individual and overall cardiovascular risk factor management with CVD events among participants with diabetes according to smoking status, as compared with participants without diabetes were analyzed.
Statistical significance was assessed using a two-sided P value of < 0.05. Statistical analyses used SAS software, version 9.4 (SAS Institute Inc).

Results
Baseline characteristics of participants according to smoking status are shown in Table 1. Of the 126,181 participants, there were 19,397 current smokers (15.4%), 6,049 former smokers (4.8%), and 100,735 never smokers (79.8%). Compared with never smokers, current smokers or former smokers had higher proportions of men, were more likely to be exposed to passive smoking in childhood, and had lower proportions of family history of CVD. Current smokers consumed fewer fruits and vegetables, were less physically active, and had higher levels of alcohol consumption than former smokers or never smokers. Former smokers had poorer glucose profiles and higher blood pressures than current smokers or never smokers.
During 455,726 person-years (median, 3.2 years; interquartile range, 3.0-4.7 years) of follow-up, we documented 3,422 CVD events: 679 participants died from cardiovascular cause, 514 had myocardial infarction, 2,055 had stroke, and 251 had heart failure; participants may have experienced more than 1 CVD event. Compared with never smokers, current smokers had significantly higher hazard of CVD events among participants  Fig. 1B). We observed similar associations of current smoking and smoking cessation with fatal and non-fatal CVD events, separately (Additional file 1: Figures S1, S2). Diabetes as well as elevated fasting glucose, elevated OGTT-2 h glucose, and elevated HbA1c were associated with increased hazards of CVD events, and these associations were more prominent among current smokers and former smokers than among never smokers ( Table 3). The multivariable-adjusted HRs for CVD events associated with diabetes were 1.61 (95% CI, 1.36-1.91) among current smokers, 1.70 (95% CI, 1.31-2.22) among former smokers, and 1.37 (95% CI, 1.26-1.49) among never smokers (P for interaction = 0.034). Significant interaction was also observed for elevated HbA1c (P for interaction = 0.011), whereas no statistically significant interaction was observed for elevated fasting glucose or elevated OGTT-2 h glucose. Similar but weaker results were observed for fatal and non-fatal CVD events separately (Additional file 1: Table S1).
Compared with participants without diabetes, participants with diabetes who had each individual controlled risk factor showed relatively lower hazards of CVD events among never smokers than among current or former smokers ( Fig. 2A). Compared with participants without diabetes, participants with diabetes who had 5 or more controlled risk factors exhibited no significantly excess risk for CVD events (HR, 1.02; 95% CI, 0.81-1.27; Table 4). Such favorable effect of 5 or more controlled risk factors on CVD risk was more prominent among never smokers (HR, 0.93; 95% CI, 0.71-1.22), but was offset among current smokers (HR, 1.28; 95% CI, 0.77-2.14) or former smokers (HR, 1.22; 95% CI, 0.66-2.28; Fig. 2B).

Discussion
In this nationwide prospective cohort study in China, we have provided novel and quantitative data about the interplay of cigarette smoking with diabetes and cardiovascular risk factor management on subsequent risk of CVD events. Compared with adults without diabetes, adults with diabetes were more susceptible to the Table 2 Association of smoking with CVD events among participants with and without diabetes 126,181 participants were included in the analysis a Adjusted for age, sex, education attainment (less than high school, high school or further education), family history of diabetes (yes, no), family history of CVD (yes, no), fruits and vegetables intake (< 4.5 cup/day, ≥ 4.5 cup/day), physical activity (active, insufficiently active, inactive), alcohol consumption, hypertension (yes, no), and dyslipidemia (yes, no) b Adjusted for CVD risk factors as listed above, as well as baseline BMI, BMI change during follow-up, and passive smoking exposure in childhood and adulthood c Interaction of diabetes with smoking status in relation to CVD events was adjusted for CVD risk factors as listed above, as well as baseline BMI, BMI change during follow-up, and passive smoking exposure in childhood and adulthood

Category
No. of participants deleterious effect of current smoking, especially heavy smoking, on CVD events. From another perspective, the excess CVD risks associated with diabetes and elevated glycemic traits were exacerbated among current smokers and former smokers, and the beneficial effect of risk factor management on CVD events among adults with diabetes seemed to be substantially attenuated by current or former smoking.
Our findings support a bidirectional interaction between smoking and diabetes on incident CVD events. Cigarette smoking and diabetes contribute independently to increased risk of CVD [4][5][6][7][8], and among people with diabetes, smoking is the key lifestyle risk factor for CVD [6]. In addition, current smokers have longer duration of exposure to passive smoking than never smokers and passive smoking is associated with an increased risk of CVD [30][31][32]. Given the considerable prevalence of smoking and the relatively low smoking cessation rate among people with diabetes and the effects of passive smoking on aspects of human health [14,15], the specific effect that smoking poses to the diabetes population deserves special attention, and exposure to secondhand smoke should be minimized or avoided in order to reduce public health hazards [6,33]. Prospective studies investigating the interaction between smoking and diabetes on CVD events are scarce and have revealed no evidence that the smoking-related CVD risk is greater in people with diabetes than in those without diabetes [34,35]. In this study, taking advantage of a large nationwide sample with detailed records of smoking behaviors and comprehensive glycemic measurements, we extended previous evidence by showing that smoking and diabetes could amplify the deleterious effect of each other on subsequent risk of CVD, and the interaction between smoking and diabetes was quite consistent when diagnosing diabetes by each of the glycemic traits, especially elevated HbA1c. Our findings thus underline the importance of avoiding or quitting smoking for the effective prevention of CVD events, particularly for people with diabetes.
Our results also support a noteworthy conclusion that the favorable effects of optimal management of cardiovascular risk factors on incident CVD events among adults with diabetes could be considerably counteracted by smoking. Previous studies have demonstrated that substantial additional reductions in major CVD events in adults with diabetes can be achieved through ideal control of cardiovascular risk factors [16,[36][37][38]. In this study, the favorable cardiovascular effect of an ideal control of 5 or more risk factors in diabetes was only seen in never smokers; the CVD risk remained high among adults with diabetes who were current smokers or former smokers, even with optimal risk factor management. These results suggest that smoking is an important barrier for people with diabetes to gain cardiovascular benefits through risk factor management.
The precise mechanisms to explain our observations are unclear. Theoretically, both smoking and diabetes contribute to the development of cardiac dysfunction via promoting oxidative stress, inducing endothelial injury, Fig. 1 Association of current smoking and smoking cessation with CVD events among participants with and without diabetes. A. Number of participants included in the analysis: 19,397 current smokers and 100,735 never smokers. Light smokers (n = 3020) referred to participants who smoked < 10 cigarettes per day, moderate smokers (n = 5051) referred to participants who smoked 10-< 20 cigarettes per day, and heavy smokers (n = 11,326) referred to participants who smoked ≥ 20 cigarettes per day. HRs (95% CIs) and P value were adjusted for age, sex, education attainment (less than high school, high school or further education), family history of diabetes (yes, no), family history of CVD (yes, no), fruits and vegetables intake (< 4.5 cup/day, ≥ 4.5 cup/day), physical activity (active, insufficiently active, inactive), alcohol consumption, hypertension (yes, no), dyslipidemia (yes, no), baseline BMI, BMI change during follow-up, and passive smoking exposure in childhood and adulthood. B. Number of participants included in the analysis: 6,049 former smokers and 19,397 current smokers. HRs (95% CIs) and P value were adjusted for the same set of variables as shown in Panel A.  126,181 participants were included in the analysis. The numbers of missing values are 120 for fasting glucose, 683 for OGTT-2 h glucose, and 179 for HbA1c a Adjusted for age, sex, education attainment (less than high school, high school or further education), family history of diabetes (yes, no), family history of CVD (yes, no), fruits and vegetables intake (< 4.5 cup/day, ≥ 4.5 cup/day), physical activity (active, insufficiently active, inactive), alcohol consumption, hypertension (yes, no), and dyslipidemia (yes, no) b Adjusted for CVD risk factors as listed above, as well as baseline BMI, BMI change during follow-up, and passive smoking exposure in childhood and adulthood c Interactions of diabetes, elevated fasting glucose, elevated OGTT-2 h glucose, and elevated HbA1c with smoking status in relation to CVD events were adjusted for CVD risk factors as listed above, as well as baseline BMI, BMI change during follow-up, and passive smoking exposure in childhood and adulthood and leading formation of atheroma [39], and these shared pathogenic pathways may underlie the interplay between smoking and diabetes on CVD. In addition, smoking is independently associated with the devastating cardiovascular health consequences for people with diabetes [39,40]. Therefore, the benefits of ideal management of other risk factors may not fully outweigh the smoking-related cardiovascular risk. Experimental studies are needed to shed light on potential mechanisms responsible for the interplay between smoking and diabetes on CVD risk.
The strengths of this study included the large nationwide sample size, the prospective study design, the detailed records of smoking behaviors and comprehensive measurements of glycemic traits, and the wellvalidated definitions of the CVD outcomes. This study also has several limitations. First, the self-reported nature of smoking information may underestimate the prevalence of smoking as ascertained by cotinine measurement, and biochemical verification is needed to confirm the findings of this study. Second, we could not fully consider the residual and unmeasured confounders, such as the use of aspirin and metformin, which has been suggested to help prevent CVD risk in certain patients [41]. Third, the large sample size of this study can enhance the reliability of our findings and has advantages in permitting stratified analyses. However, because of the large sample size in this study, a statistically significant finding might not necessarily be clinically significant; thus, significant findings derived from this study must be analyzed and interpreted with caution. Fourth, similar to previous National Smoking Survey [42] and China Noncommunicable Disease Surveillance [43], the proportion of smokers were extremely low in Chinese women. Therefore, the limited number of smokers restricted the statistical power to perform sex-stratification analysis in this study. Fifth, the participants were middle-aged and elderly Chinese Fig. 2 Association of cardiovascular risk factor management with CVD events among participants with diabetes according to smoking status, as compared with participants without diabetes. A. Number of participants included in the analysis: 96,981 without diabetes, 12,408 with diabetes and fruits and vegetables intake ≥ 4.5 cup/ day, 19,921 with diabetes and physical activity ≥ 600 MET-min/ week; 9980 with diabetes and HbA1c < 6.5%, 8052 with diabetes and SBP/DBP < 130/80 mmHg, 10,312 with diabetes and LDL cholesterol < 100 mg/dL, and 28,032 with diabetes and no CKD. HRs (95% CIs) were adjusted for age, sex, education attainment (less than high school, high school or further education), family history of diabetes (yes, no), family history of CVD (yes, no), fruits and vegetables intake (< 4.5 cup/day, ≥ 4.5 cup/day; not for fruits and vegetables intake analysis), physical activity (active, insufficiently active, inactive; not for physical activity analysis), alcohol consumption, hypertension (yes, no; not for SBP/DBP analysis), dyslipidemia (yes, no; not for LDL cholesterol analysis), baseline BMI, BMI change during follow-up, and passive smoking exposure in childhood and adulthood. B. Number of participants included in the analysis: 96,981 without diabetes, 9753 with diabetes and ≤ 2 controlled risk factors, 9667 with diabetes and 3 controlled risk factors, 6,846 with diabetes and 4 controlled risk factors, and 2934 with diabetes and ≥ 5 controlled risk factors. HRs (95% CIs) were adjusted for age, sex, education attainment (less than high school, high school or further education), family history of diabetes (yes, no), family history of CVD (yes, no), alcohol consumption, baseline BMI, BMI change during follow-up, and passive smoking exposure in childhood and adulthood. ▸ adults, and a majority of current smokers and former smokers were men. Thus, the generalizability of our findings to other groups should be cautious.

Conclusions
Smoking and diabetes significantly interacted with each other in imposing increased risks of subsequent CVD events, and the cardiovascular benefits of risk factor management in diabetes could be counteracted by former or current smoking. Our findings highlight the particular importance of avoiding or quitting smoking to reduce the preventable burden of CVD in people with diabetes.
Additional file 1: Table S1. Association of diabetes, elevated fasting glucose, elevated OGTT-2h glucose, and elevated HbA1c with non-fatal CVD events and CVD mortality among current, former, and never smokers. 126,181 participants were included in the analysis. The numbers of missing values are 120 for fasting glucose, 683 for OGTT-2h glucose, and 179 for HbA1c. aAdjusted for age, sex, education attainment (less than high school, high school or further education), family history of diabetes (yes, no), family history of CVD (yes, no), fruits and vegetables intake (<4.5 cup/ day, ≥4.5 cup/day), physical activity (active, insufficiently active, inactive), alcohol consumption, hypertension (yes, no), dyslipidemia (yes, no), baseline BMI, BMI change during follow-up, and passive smoking exposure in childhood and adulthood. bPerson-years for CVD mortality was longer than that for non-fatal CVD events, because participants may have experienced more than one CVD event. Figure S1. Association of current smoking with non-fatal CVD events and CVD mortality among participants with and without diabetes. Number of participants included in the analysis: 19,397 current smokers and 100,735 never smokers. Light smokers (n=3,020) referred to participants who smoked <10 cigarettes per day, moderate smokers (n=5,051) referred to participants who smoked 10-<20 cigarettes per day, and heavy smokers (n=11,326) referred to participants who smoked ≥20 cigarettes per day. HRs (95% CIs) and P value were adjusted for age, sex, education attainment (less than high school, high school or further education), family history of diabetes (yes, no), family history of CVD (yes, no), fruits and vegetables intake (<4.5 cup/day, ≥4.5 cup/day), physical activity (active, insufficiently active, inactive), alcohol consumption, hypertension (yes, no), dyslipidemia (yes, no), baseline BMI, BMI change during follow-up, and passive smoking exposure in childhood and adulthood. Figure S2. Association of smoking cessation with nonfatal CVD events and CVD mortality among participants with and without diabetes. Number of participants included in the analysis: 6,049 former smokers and 19,397 current smokers. HRs (95% CIs) and P value were adjusted for age, sex, education attainment (less than high school, high school or further education), family history of diabetes (yes, no), family history of CVD (yes, no), fruits and vegetables intake (<4.5 cup/day, ≥4.5 cup/day), physical activity (active, insufficiently active, inactive), alcohol consumption, hypertension (yes, no), dyslipidemia (yes, no), baseline BMI, BMI change during follow-up, and passive smoking exposure in childhood and adulthood.