This retrospective cohort study used National Health Insurance System-National Sample Cohort (NHIS-NSC) database established by the National Health Insurance System in South Korea. NHIS-NSC database contains sociodemographic characteristics and medical information of about one million Koreans (26). A sample cohort included in NHIS-NSC database was sampled based on a stratified random sampling method, and followed up to 2019 (27). It includes qualification data, medical records such as diagnoses based on the International Statistical Classification of Diseases and Related Health Problems 10th revision (ICD-10) or medical procedures, and drug prescriptions. This study used NHIS-NSC dataset from 2002 to 2019.

Among 1,021,208 participants in NHIS-NSC dataset, we first selected participants who underwent medical health examinations at least once between 2008 and 2011 considering that data on waist circumference has been available since 2008 (28). Among 466,528 participants who received health examinations at least one time between 2008 and 2011, we excluded 177,240 participants who had only one time of medical check-up between 2008 and 2011, since at least two time of examination records is needed to capture a participant’s change of waist circumference, weight and smoking status. During 4-year of screening period, the first health examination date was defined as a participant’s first screening date, and the last health examination date was defined as a participant’s follow-up screening date. Next, we excluded 20,492 participants with missing data for at least one of measurements used in this study. To minimize possibility of reverse causality between waist or weight change after smoking cessation and the incidence of CVD or all-cause death, 18,581 participants who were diagnosed with CVD before follow-up screening date were also excluded. Lastly, we only included participants with type 2 diabetes, who satisfy any of following criteria: (i) fasting serum glucose concentration ≥ 126 mg/dL, or (ii) at least two claims of ICD-10 code (E11-E14) with the prescription of anti-diabetic agents before follow-up screening date (29). Finally, 32,142 participants were included in this study, as represented in Figure 1 .

Study outcome was the incidence of CVD or all-cause death. The incidence of CVD was defined as follows: (i) at least two claims of ICD-10 code (I20–I25) for ischemic heart disease, or (ii) at least two claims of ICD-10 (I60–I64) for cerebrovascular disease after follow-up screening date (30). The occurrence of all-cause death was defined based on the presence of a participant’s date of death caused by any reason. Follow-up end date for CVD or all-cause death was defined as the earliest date of following three dates: (i) first incidence date of CVD (ii) date of death, or (iii) study end date (December 31, 2019).

Based on change from a patient’s self-reported smoking status at first screening date to that at follow-up screening date, participants were categorized into four groups as follows: (i) change from any none/former/current smoker to current smoker (current smoker), (ii) change from current smoker to none/former smoker, or change from non-smoker to former smoker (recent quitter), (iii) change from former smoker to none/former smoker (long-term quitter), and (iv) change from non-smoker to non-smoker (non-smoker) (24).

Waist circumference was measured as the halfway between the lowest rib cage and top of the iliac crest in an upright position, and recorded in millimeter (31). Based on absolute change in waist circumference, defined as waist circumference at first screening date subtracted from that at follow-up screening date, participants were categorized into three groups as follows: <0.1 cm (no waist gain), 0.1–5.0 cm, and ≥5.0 cm. Cut-off points for waist circumference were set based on commonly used cut-off points, or incremental units to segment participants according to waist circumference in previous studies (32–35). Weight was measured using an electronic scale while a participant was dressed in simple clothes, and recorded as a valid integer value in kilograms (36). In line with waist circumference, change in weight was computed as weight at follow-up screening date minus that at first screening date, and utilized to categorize participants into three groups as follows: <2 kg (no weight gain), 2–5 kg, and ≥5 kg. Cut-off points for weight was determined based on previous studies on smoking or weight among Asian population (37, 38).

In order to investigate the risk for CVD or all-cause death in association with waist gain after smoking cessation, participants were classified into one of eight groups as follows: current smoker, recent quitter with no waist gain, recent quitter with waist gain of 0.1-5.0 cm, recent quitter with waist gain of ≥5.0 cm, long-term quitter with no waist gain, long-term quitter with waist gain of 0.1-5.0 cm, long-term quitter with waist gain of ≥5.0 cm, and non-smoker. For the purpose of exploring the risk for CVD or all-cause death in association with weight gain after smoking cessation, participants were classified into one of eight groups as follows: current smoker, recent quitter with no weight gain, recent quitter with weight gain of 2-5 kg, recent quitter with weight gain of ≥5 kg, long-term quitter with no weight gain, long-term quitter with weight gain of 2-5 kg, long-term quitter with weight gain of ≥5 kg, and non-smoker.

To minimize confounding effects by nuisance variables, we considered following variables as covariates for adjustment: age, sex, household income level, systolic blood pressure, diastolic blood pressure, total cholesterol, alcohol consumption, physical activity, comorbidities at follow-up screening date, and body mass index (BMI) at first screening date. Household income level was categorized as follows: <40^th percentile (low), 40^th-80^th percentile (middle), and 80^th-100^th percentile (high). BMI was calculated as a participant’s weight in kilograms divided by the square of the participant’s height in meters, and categorized into following five groups according to World Health Organization’s classification guide for Asians: <18.5 kg/m² (underweight), 18.5–23.0 kg/m² (normal weight), 23.0–25.0 kg/m² (overweight), 25.0–30.0 kg/m² (class I obese), and ≥30 kg/m² (class II obese) (39). Total cholesterol was measured by the enzymatic method after drawing an examinee’s blood with a disposable syringe or vacuum test tube including a disposable needle when the examinee was in fasting state for at least 12 hours before examination (40). Alcohol consumption was classified into three groups according to the average number of drinking days weekly as follows: <1 (none), 1–4 (light-to-moderate), ≥5 (heavy). Physical activity was classified into three groups according to the average number of days of exercise weekly as follows: <1 (none), 1–4 (light-to-moderate), ≥5 (vigorous).

We investigated whether a participant has history of hypertension, dyslipidemia, or cancer, which are representative comorbidities among patients with type 2 diabetes (41). Hypertension was defined according to the following criteria: (i) systolic/diastolic blood pressure ≥ 140/90 mmHg, or (ii) at least two claims of ICD-10 code (I10-I15) with the prescription of anti-hypertensive agents. Dyslipidemia was defined according to the following criteria: (i) total cholesterol level ≥ 240 mg/dL, or (ii) at least two claims of ICD-10 code (E78) with the prescription of lipid-lowering agents. Cancer was defined as using at least two claims of ICD-10 code (C00–C97).

Descriptive statistics were calculated for numerical or categorical variables included in this study. While numerical variables were summarized as mean with standard deviation, categorical variables were summarized as number of cases with proportion in percentage. In order to find statistical difference in a variable among groups, ANOVA test was used for numerical variables, and Chi-square test was used for categorical variables. We calculated hazard ratio (HR) and corresponding 95% confidence interval (CI) using Cox proportional hazards model (42) to compare the risk of CVD or all-cause death among different smoking cessation groups, or among different smoking groups in combination with post-cessation weight or waist change. Unadjusted HR was calculated from univariate Cox model. Adjusted HR was calculated from hierarchically adjusted Cox models as follows: (i) Model 1 (sex and age adjusted), (ii) Model 2 (Model 1+baseline BMI, household income level, systolic blood pressure, diastolic blood pressure, total cholesterol, alcohol consumption, physical activity, hypertension, dyslipidemia, cancer adjusted). The proportionality assumption in hazards over time was checked by score test, which proved the appropriateness of this assumption in result. All analysis was performed using SAS software (version 9.4, SAS Institute, Cary, NC, USA) and R version 4.2.3 (R Foundation for Statistical Computing, Vienna, Austria; http://www.R-project.org). A two-sided p-value of less than 0.05 was considered statistically significant.

Of 32,142 eligible participants, 5,845 participants were newly diagnosed with CVD during a median follow-up of 8.42 years (interquartile range: 8.00–9.00), and 3,723 participants died during a median follow-up of 8.17 years (interquartile range: 8.17–9.08). Table 1 represents the baseline characteristics of study participants according to change in smoking status. In terms of all variables in Table 1 , participants tended to have significantly different mean or proportion according to change in smoking status. Compared to current smokers, recent or long-term quitters were older, had higher household income, were more likely to be overweight at baseline, drank less, worked out more frequently in a week, and were more likely to have a history of comorbidities. Both systolic and diastolic blood pressure were higher in recent or long-term quitters than those for current smokers.

Table 2 represents HR and 95% CI for the incidence of CVD or all-cause death associated with change in smoking status. After adjusting for possible confounding factors, HR for CVD were 0.94 (95% CI: 0.85–1.03; p = 0.192), 0.82 (95% CI: 0.74–0.90; p <.001), and 0.82 (95% CI: 0.75–0.90; p <.001) for recent quitters, long-term quitters, non-smokers compared to current smokers, respectively. Long-term quitters or non-smokers had significantly lower risk for CVD, compared to current smokers. In case of all-cause death, fully adjusted HR for all-cause death was 0.67 (95% CI: 0.60–0.75; p <.001), 0.62 (95% CI: 0.56–0.70; p <.001), and 0.55 (95% CI: 0.50–0.61; p <.001) for recent quitters, long-term quitters, non-smokers compared to current smokers, respectively. Compared to current smokers, the risk for all-cause death was significantly lower in all of three groups (recent quitters, long-term quitters, non-smokers).

Figure 2 graphically shows HR and 95% CI for the incidence of CVD or all-cause death associated with waist gain after smoking cessation, and corresponding results are also described in Supplementary Table 1 . In case of recent quitters, fully adjusted HR for CVD was 0.98 (95% CI: 0.87–1.10; p = 0.683), 0.85 (95% CI: 0.73–1.00; p = 0.053), and 0.95 (95% CI: 0.80–1.14; p = 0.589) for recent quitters with no waist gain, recent quitters with waist gain of 0.1-5.0 cm, and recent quitters with waist gain of ≥5.0 cm, compared to current smokers, respectively. In case of long-term quitters, fully adjusted HR for CVD was 0.88 (95% CI: 0.78–0.99; p = 0.032), 0.68 (95% CI: 0.57–0.81; p <.001), and 0.82 (95% CI: 0.67–1.00; p = 0.051) for long-term quitters with no waist gain, long-term quitters with waist gain of 0.1-5.0 cm, and long-term quitters with waist gain of ≥5.0 cm, compared to current smokers, respectively. Compared to current smokers, HR for CVD was 0.82 (95% CI: 0.75–0.90; p <.001) for non-smokers.

In terms of all-cause death, fully adjusted HR for all-cause death was 0.69 (95% CI: 0.60–0.79; p <.001), 0.62 (95% CI: 0.52–0.74; p <.001), and 0.70 (95% CI: 0.57–0.86; p <.001) for recent quitters with no waist gain, recent quitters with waist gain of 0.1-5.0 cm, and recent quitters with waist gain of ≥5.0 cm, compared to current smokers, respectively. In case of long-term quitters, fully adjusted HR for all-cause death was 0.56 (95% CI: 0.49–0.65; p <.001), 0.69 (95% CI: 0.57–0.83; p <.001), and 0.75 (95% CI: 0.60–0.93; p = 0.009) for long-term quitters with no waist gain, long-term quitters with waist gain of 0.1-5.0 cm, and long-term quitters with waist gain of ≥5.0 cm, compared to current smokers, respectively. Compared to current smokers, HR for all-cause death was 0.55 (95% CI: 0.50–0.61; p <.001) for non-smokers.

Figure 3 is the forest plot representing HR and 95% CI for the incidence of CVD or all-cause death associated with weight gain after smoking cessation, and corresponding results are also described in Supplementary Table 2 . In case of recent quitters, fully adjusted HR for CVD was 0.93 (95% CI: 0.83–1.05; p = 0.245), 0.93 (95% CI: 0.81–1.07; p = 0.329), and 0.99 (95% CI: 0.75–1.32; p = 0.955) for recent quitters with no weight gain, recent quitters with weight gain of 2-5 kg, and recent quitters with weight gain of ≥5 kg, compared to current smokers, respectively. In case of long-term quitters, fully adjusted HR for CVD was 0.79 (95% CI: 0.71–0.89; p <.001), 0.85 (95% CI: 0.74–0.98; p = 0.029), and 0.84 (95% CI: 0.60–1.17; p = 0.296) for long-term quitters with no weight gain, long-term quitters with weight gain of 2-5 kg, and long-term quitters with weight gain of ≥5 kg, compared to current smokers, respectively. Compared to current smokers, HR for CVD was 0.82 (95% CI: 0.75–0.90; p <.001) for non-smokers.

In terms of all-cause death, fully adjusted HR for all-cause death was 0.66 (95% CI: 0.58–0.75; p <.001), 0.65 (95% CI: 0.55–0.77; p <.001), and 0.90 (95% CI: 0.64–1.25; p = 0.509) for recent quitters with no weight gain, recent quitters with weight gain of 2-5 kg, and recent quitters with weight gain of ≥5 kg, compared to current smokers, respectively. In case of long-term quitters, fully adjusted HR for all-cause death was 0.58 (95% CI: 0.50–0.66; p <.001), 0.67 (95% CI: 0.56–0.79; p <.001), and 0.97 (95% CI: 0.70–1.33; p = 0.834) for long-term quitters with no weight gain, long-term quitters with weight gain of 2-5 kg, and long-term quitters with weight gain of ≥5 kg, compared to current smokers, respectively. Compared to current smokers, HR for all-cause death was 0.55 (95% CI: 0.50–0.61; p <.001) for non-smokers.