The interaction between diet quality and cigarette smoking on the incidence of hypertension, stroke, cardiovascular diseases, and all-cause mortality

This study aimed to examine the interaction between diet quality indices (DQIs) and smoking on the incidence of hypertension (HTN), stroke, cardiovascular diseases, and all-cause mortality. We prospectively followed 5720 participants and collected dietary data via a validated food frequency questionnaire to calculate DQI-international (DQI-I) and DQI-revised (DQI-R). Considering an interaction analysis, we classified participants based on diet quality (median: higher/lower) and smoking status. Over 9 years of follow-up, higher diet quality scores were associated with a lower risk of stroke and mortality. While current smokers had a higher risk of stroke and mortality but had a lower risk of developing HTN. Compared to the current smokers with lower diet quality, nonsmokers with higher diet quality according to the DQI-I [HR 0.24; 95% CI (0.08, 0.66)], and DQI-R [HR 0.20; 95% CI (0.07, 0.57)] had a lower risk of stroke. Moreover, the lower risk of mortality was more evident in nonsmokers with higher DQI-I [HR 0.40; 95% CI (0.22–0.75)] and DQI-R scores [HR 0.34; 95% CI (0.18–0.63)] compared to nonsmokers with lower diet quality. While higher DQI-I and DQI-R scores were associated with a lower risk of stroke and mortality, this beneficial effect may be negated by smoking.


Results
The mean age ± SD of the participants recruited for the study of cardiovascular events was 46.42 years, and 44.8% were men.The baseline characteristics of the study participants based on diet quality and smoking status are reported in Table 1.In the total population, the mean ± SD of DQI-I and DQI-R were 63.41 ± 7.98 and 71.20 ± 12.61, respectively.Based on both DQI-I and DQI-R, individuals with higher vs. lower diet quality showed a higher proportion of men, and had higher mean age, education level, FBS, TG, SBP levels, and energy intake.Individuals with higher DQI-I had lower levels of HDL-C and employment rate.Also, those with higher DQI-R had higher BMI and higher physical activity levels compared with those with lower DQI-R.Compared to non-smokers, current smokers had significantly a higher percentage of men and TG levels, while mean age, BMI, physical activity, and HDL-C levels were significantly lower in the current smokers compared to nonsmokers.Moreover, the energy intake was higher, and both quality diet scores were lower for current smokers than non-smokers.
Supplementary Table S1 displays the baseline characteristics of the study participants recruited for the study of HTN.The mean age ± SD of the participant's age was 38.25 years, and 42.6% were men.The total population Table 1.Baseline characteristics of the study participants for cardiovascular disease incidents and all-cause mortality based on diet quality and smoking status (n = 5048).BMI body mass index, PA physical activity, FBS fasting blood sugar, TC total cholesterol, TG triglycerides, HDL-C high-density lipoprotein cholesterol, SBP systolic blood pressure, FHCVD family history of premature CVD, DQI-I diet quality index-international, DQI-R diet quality index-revised.Values are reported as mean ± standard deviation for normally distributed variables, median (interquartile range) for non-normally distributed variables, and percentage for categorical variables.P-values were obtained by independent sample t-test for normally distributed continuous variables, the Mann-Whitney U test for non-normally distributed variables, and chi-square for categorical variables.had the mean ± SD of DQI-I and DQI-R of 62.43 ± 8.03 and 70.00 ± 12.81, respectively.Considering the DQI-I and DQI-R, participants with higher diet quality, had a higher percentage of men, higher mean age and BMI, higher physical activity and education level, higher energy intake, and exhibited higher levels of FBS, TC, TG, and SBP.According to the DQI-I, participants with higher diet quality had a lower HDL-C level.Compared to non-smokers, current smokers were more likely to be male, had a higher age, higher physical activity level, higher levels of TC, TG, SBP, and higher energy intake, however, they had lower HDL-C, lower scores of DQI-I and its components including variety and moderation, and also lower DQI-R score and its moderation score.After a mean follow-up of 9.0 years, 357 cases of CVDs (7.1%), 54 cases of stroke (1.1%), and 131 cases of mortality (2.6%) were ascertained.Also, considering the HTN outcome, 1032 cases of HTN (18%) were recorded after a mean follow-up period of 7.8 years.Our results indicated the robust accuracy of the multivariable models in predicting the incidence of the investigated outcomes.The Harrel's C index scores for the associations between smoking and HTN (0.792), smoking and stroke (0.880), smoking and CVDs (0.823), smoking and mortality (0.847), DQI-I and HTN (0.791), DQI-I and stroke (0.895), DQI-I and CVDs (0.823), DQI-I and mortality (0.846), DQI-R and HTN (0.791), DQI-R and stroke (0.895), DQI-R and CVDs (0.823), and DQI-R and mortality (0.849) signify their predictive capability.Considering the interaction between diet quality and smoking, the multivariable models exhibited good predictive power concerning the incidence of HTN (DQI-I = 0.791, DQI-R = 0.791), stroke (DQI-I = 0.887, DQI-R = 0.887), CVDs (DQI-I = 0.823, DQI-R = 0.823), and all-cause mortality (DQI-I = 0.849, DQI-R = 0.851).
Table 2 represents the HR and 95% CI for cardiovascular events, all-cause mortality, and HTN across quartiles of DQIs.The higher DQI-I and DQI-R scores were associated with a higher risk of total cardiovascular events in the crude model but after accounting for covariates in model 1 and model 2, the associations became nonsignificant.The hazard of stroke decreased significantly across quartiles of DQI-I and DQI-R in both adjusted models.Compared with those in the first quartile of DQI-I and DQI-R, the fully adjusted HRs (95% CIs) were 0.37 (0.17, 0.82) and 0.33 (0.15, 0.72), respectively.The hazard of the all-cause mortality also decreased across the quartiles of DQI-I and DQI-R in the adjusted models; the fully adjusted HR (95% CI) in those in the highest quartile compared to first quartile was 0.57 (0.34, 0.97) for DQI-I and 0.51 (0.30, 0.86) for DQI-R.The hazard of HTN increased significantly across quartiles DQI-I (P trend = 0.001) and DQI-R (P trend = 0.003) in the unadjusted model but by inclusion of covariates in models 1 and 2 the associations became non-significant.
Figure 1 indicates the hazard ratio for incidence of CVDs, stroke, mortality, and HTN in current smokers compared with non-smokers.In the crude model, the risk of all-cause mortality was significantly higher in the current smokers than non-smokers [HR 1.87, 95% CI (1.19-2.93)]; the association remained significant in the final adjusted model [HR 2.11; 95% CI (1.30-3.45)].A significantly increased risk of stroke was also observed in the current smokers than the non-smokers after accounting for all potential covariates [HR 2.32; 95% CI (1.06-5.10)].The risk of HTN was significantly lower in the current smokers compared to non-smokers in the fully adjusted model [HR 0.77; 95% CI (0.62-0.96)].
Table 3 displays the HR (95% CI) for cardiovascular events, all-cause mortality, and HTN based on the combined effects of diet quality and smoking status.No significant interaction (P interaction > 0.05) was noted among the studied groups concerning cardiovascular events, all-cause mortality, and HTN.Compared to current smokers with poor diet quality, the hazard (95% CI) of stroke was 0.23 (0.08, 0.63) in adjusted model 1 and 0.24 (0.08, 0.66) in the adjusted model 2 in non-smokers with good diet quality based on DQI-I.A lower risk of stroke was also observed in non-smokers with good diet quality based on DQI-R compared to current smokers with poor diet quality in the adjusted models.The risk of all-cause mortality was significantly lower in both non-smokers' groups with higher or lower diet quality compared to the current smokers with poor diet quality in the crude model and after adjusting for potential covariates.The hazard of HTN was not significantly different across the combined groups.

Discussion
The current study revealed that a high-quality diet was linearly associated with a lower risk of stroke and mortality.Compared to non-smokers, current smokers had an higher risk of stroke, CVDs, and mortality, and a lower risk of HTN.Additionally, non-smokers with higher diet quality had a lower risk of stroke compared to current smokers with lower diet quality.Non-smokers experience a lower risk of mortality than current smokers with lower diet quality.Notably, the lower risk for stroke and all-cause mortality was more pronounced in non-smokers with higher diet quality than in those with lower diet quality.
The similarities in the associations between the DQI-I and the DQI-R with the studied outcomes are likely attributable to similarities in their food and nutrient components.In general, higher DQI-I and the DQI-R scores indicate a healthier diet, characterized by a variety of food choices, adequate intake of fruits, vegetables, grains, and micronutrients, and suitable percentages of total energy intake for total fat, saturated fats, and high-calorie foods.Additionally, the DQI-R considers moderation of alcohol consumption, which makes it valuable in populations where alcohol use is either excessive or contraindicated.Although both indices evaluate an approximately equal number of dietary components, the DQI-I encompasses a broader spectrum of dietary aspects compared to the DQI-R.In contrast to the DQI-R, the DQI-I assesses the adequacy of fiber, protein, and vitamin C intake, as well as the balance in ratios between macronutrients and fatty acids.
Despite the limited number of studies examining the link between diet quality, as assessed by the DQIs, and the incidence of CVDs and all-cause mortality, our findings contribute additional evidence supporting the role of overall diet quality in the prevention of CVDs outcomes.In a study conducted within the framework of the Korean National Health and Nutrition Examination Surveys, it was observed that, based on DQI-I, individuals with CVDs exhibited a lower diet quality in comparison to those without CVDs 24  www.nature.com/scientificreports/hypertriglyceridemia, and reduced HDL-C levels among obese adults 25 .A cross-sectional study conducted by Shiraseb et al. revealed an inverse association between the DQI-I score and serum inflammatory biomarkers 19 .However, a cross-sectional study by Daneshzad et al. revealed no significant association between DQI-I score and cardiovascular disease risk factors in type 2 diabetic women, possibly due to low statistical power.Of note, it has been suggested that different ratings of DQIs for evaluating diet quality in different populations, such as diabetic patients, be created 26 .
Current smokers seem to exhibit a lower risk of HTN compared to non-smokers.Previous studies suggest that the lower risk of HTN in current smokers is attributed to higher levels of physical activity and lower weight of smokers.Khodamradi et al., following a 20-year follow-up, implied that the lower risk of HTN among smokers may be linked to their increased participation in work-related activities 21 .Besides, smoking can impact the hypothalamus, leading to appetite suppression, reduced dietary intake, and elevated catecholamine levels.This results in higher energy consumption in peripheral tissues and, ultimately, a lower body weight for smokers 27 .Hazard ratio (95% confidence intervals) for cardiovascular events, all-cause mortality, and hypertension based on the quartiles of dietary quality indices.DQI-I diet quality index-international, DQI-R diet quality index-revised.Model 1, was adjusted for sex, age, systolic blood pressure, fasting blood sugar, serum cholesterol, serum triglyceride, high-density lipoprotein cholesterol, energy intake, BMI, and physical activity.Model 2 additionally adjusted for medication use, employment status, education level, and family history of premature CVD.All diet quality indices were constructed according to each 1000 kcal of energy intake.www.nature.com/scientificreports/However, smoking causes a temporary increase in blood pressure, which returns to normal levels after 20 min of smoking 10 .According to data from the CARDIA (Coronary Artery Risk Development in Young Adults) study, the detrimental effects of smoking on blood pressure changes and pulse pressure (as calculated by subtracting DBP from SBP) are more pronounced than those of blood pressure levels alone.Notably, cigarette smoking can cause stiffness in large elastic arteries, like the aorta.This stiffness disrupts the ability of vessels to adjust to pressure changes during systole and diastole.Although there was no significant difference between blood pressure levels in current and non-smokers, the acute increase in blood pressure coupled with the progression of atherosclerosis may amplify the likelihood of unfavorable outcomes for smokers 28 .
The risk of stroke in comparison to other CVDs events, was significantly higher among current smokers.Findings from the Atherosclerosis Risk in Communities (ARIC) study indicate relative risks for lacunar and cardioembolic strokes in current smokers of 2.30 and 1.94, respectively 29 .In addition, Kawachi et al. demonstrated that, within the framework of the Nurse's Health Study, nurses who smoke are almost three times more likely to experience a stroke 30 .The elevated risk associated with smoking and stroke is attributed to factors like increased atherosclerotic plaque formation, heightened thrombosis, reduced cerebral perfusion, facilitation of aneurysm formation and rupture, structural damage to the arterial wall, carboxyhemoglobinemia, heightened platelet aggregability, and elevated fibrinogen levels 31 .Additionally, in animal models, a constituent of environmental tobacco smoke like 1,3-butadiene, can accelerate atherosclerosis, further linking smoking to CVDs risk 10 .
Previous studies have presented contrasting perspectives regarding the interaction effects of diet and smoking on health outcomes.One perspective suggests that, due to smokers' weakened nutritional and clinical status, a healthy diet is expected to have more pronounced effects on smokers than non-smokers 32 .In addition, in alignment with our study's results, smoking emerged as an effect modifier in the association between diet quality and NCDs.In other words, smoking might diminish the potential beneficial impact of a high-quality diet.Geng et al. reported an increased risk of CVDs mortality in individuals who were genetically prone to smoking and who also had lower DASH (dietary approach to stop hypertension) scores 7 .Moreover, Margetts et al. suggested that the risk of oxidative damage is elevated in smokers due to a combination of insufficient nutrient intake and the impact of smoking on nutrient metabolism 33 .Of note, smoking and a low-quality diet contribute to CVDs development through shared pathways like endothelial dysfunction and inflammation 34 .While increased TG and decreased HDL-C levels are the main characteristics of metabolic syndrome in smokers 35 , previous studies have established a direct association between adherence to a high-quality diet, assessed by the DQI-I, and higher serum HDL-C levels 20 .A recent study by Noruzzadeh et al. suggests that among current smokers, individuals with higher diet quality experience a lower risk of all-cause mortality compared to current smokers with lower diet quality 36 .Additionally, the study found that this lower risk associated with higher diet quality was more pronounced among former smokers 36 .Although some studies emphasize the essential role of nutrition in overall health, they suggest that diet may not influence the relationship between smoking and disease.Li and colleagues suggested that the severity of asthma symptoms in smokers was not influenced by the quality of their diet 37 .In contrast, a high-quality diet, as measured by the Alternative Healthy Eating Index 2010, was linked to a decrease in the severity of asthma symptoms in nonsmokers 37 .Millen et al. suggested that smokers who adopt heart-healthy dietary modifications may have certain advantages, but their risk of CVDs and all-cause mortality remains higher compared to non-smokers who also follow a heart-healthy diet 6 .Of note, DQIs are a measure of overall dietary quality in the general population, but considering the specific health condition of smokers, different ratings of DQIs or other dietary indices designed to correct these conditions may be more suitable in preventing CVDs incidence and all-cause mortality.It is crucial to acknowledge that our results may be susceptible to residual confounding from smoking-related factors, including smoking intensity and duration.
This study has notable strengths, including an extended follow-up duration, an investigation of the combined impact of diet quality and smoking, the use of validated and reproducible FFQ to collect dietary data, and control for various potentially confounding variables.Nevertheless, this study has considerable limitations.Like all observational studies, it may be susceptible to measurement bias.Additionally, due to the lack of available data, alcohol consumption was not incorporated as a food component in the computation of the DQI-R.Last, as an observational study, it cannot fully eliminate the impact of all potential residual confounding variables.
Based on our findings, it appears that smoking may counteract the benefits of a high-quality diet within the general population.Furthermore, existing literature suggests that the dietary habits of current smokers could influence their mortality risk.Given the limitations of this study, future studies could explore the interaction between diet quality and smoking across diverse populations.Employing alternative interaction indices like the relative excess risk due to interaction (RERI) and synergy index (SI) may provide deeper insights into their combined impact on health outcomes.Furthermore, employing restricted cubic splines (RCS) analysis to assess the relationship between smoking intensity (measured by the number of cigarettes per day) and duration (packyears) and health-related outcomes could yield more precise estimates of the impact of smoking on health risks.Given the observed interaction between higher diet quality and smoking regarding health outcomes, future nutrition studies should acknowledge smoking as a significant confounding variable.Also, considering the detrimental impact of smoking on overall mortality and the incidence of CVDs, strategies such as smoking cessation programs, smoking reduction initiatives, and the implementation of smoking bans in public areas can prove effective in alleviating the smoking-related burden of diseases.
In conclusion, following a high-quality diet can be associated with a lower risk of stroke and all-cause mortality, however, these beneficial effects may be eliminated by the adverse impact of smoking.Further research in different populations is warranted to examine the individual and combined effects of dietary quality indices and smoking on the risk of NCDs.

Study population
The current investigation was carried out within the Tehran Lipid and Glucose Study (TLGS) outline.The TLGS is an ongoing population-based research initiative designed to identify risk factors for NCDs and advocate for healthier lifestyles.Since 1999, six examinations, involving 15,005 participants who undergo routine physical examinations, laboratory tests, and updates to their medical histories every three years, have been conducted 38 .
In the third examination of the TLGS (2006-08), 3686 individuals were chosen at random for dietary assessment.In the subsequent fourth examination (2009-2011), 7956 individuals were randomly selected and participated in the dietary assessment.For the present study, participants with complete dietary intake data from the third examination and those newly enrolled in the fourth examination were considered.Notably, participants from the third examination who exhibited under-reporting or over-reporting of energy intake (≤ 800 kcal/day and ≥ 4200 kcal/day, respectively) were excluded (n = 233).However, if their energy intake fell within the normal range at the fourth examination (n = 100), their dietary data from the fourth examination were considered baseline dietary data.After excluding individuals with over-or under-reports of energy intake in the fourth examination (n = 502), a total of 8914 participants with complete dietary intake data were included.
For the HTN outcome, exclusions were made for individuals aged < 18 years (n = 1270), individuals with any type of CVDs (n = 408), history of cancer (n = 35), pregnant or lactating women (n = 154), individuals with missing data on HTN status (n = 52), and those who were hypertensive at baseline (n = 1276).It is worth noting that some participants may belong to more than one exclusion category.Additionally, 299 participants were lost to follow-up, resulting in 5720 participants remaining for the final analysis of the HTN incident (Fig. 2).
Concerning the incidence of CVDs, and all-cause mortality, among the 5532 individuals aged ≥ 30 years, exclusions were made for those with any type of CVD (n = 383), a history of cancer (n = 32), pregnancy or lactation (n = 70) at the baseline of the study.Of note, some participants may belong to more than one exclusion category.Ultimately, a total of 5048 participants, free of CVDs were included in the study (Fig. 2) and were followed up until March 20, 2018, for the analysis of CVDs, and all-cause mortality.

Dietary assessment
Nutrition specialists collected dietary intake data by employing a validated semi-quantitative food frequency questionnaire (FFQ) comprising 168 items 39 .This questionnaire assessed the frequency of participants' food consumption over the preceding year, categorizing responses into daily, weekly, or monthly frequencies.To ensure accuracy and consistency in the data analysis, portion sizes originally expressed in household measurements were converted into grams.The nutritional values of the food items were determined using the United In the case of all-cause mortality, an authorized local physician collected information from hospital records or death certificates.A committee, comprising key professionals including a chief researcher, an internist, an endocrinologist, a cardiologist, an epidemiologist, and the individual responsible for compiling outcome data, evaluated the results obtained.

Statistical analyses
The statistical analysis was performed utilizing IBM SPSS Statistics (Version 26.0,IBM Corp., Armonk, NY, USA).The characteristics of the study population are reported as percentages for categorical variables, means ± standard deviation (SDs) for normally distributed variables, and medians (interquartile ranges) for non-normally distributed variables.Comparisons between the groups, including those with higher and lower diet quality or current and non-smokers, were performed using the chi-square test for categorical variables, independent sample t-test for normally distributed continuous variables, and the Mann-Whitney U test for non-normally distributed variables.Our study did not encounter missing data.
Cox proportional hazard regression was employed to explore the associations of DQI quartiles and smoking status (non-smokers vs current smokers) with the incidence of HTN, stroke, CVDs, and all-cause mortality.The duration of follow-up was calculated from the enrollment date in the study until the occurrence of the first CVDs or mortality event or until the last follow-up date.Additionally, for the incidence of HTN, the follow-up duration was defined as the midpoint between the date of the follow-up visit where HTN was initially identified and the last follow-up visit preceding the diagnosis.
Additionally, an interaction analysis was conducted based on the joint classification of DQIs (median; higher/ lower) and smoking status (current smokers/non-smokers).For the interaction analysis, four groups were created: 1) current smokers with lower diet quality, 2) current smokers with higher diet quality, 3) non-smokers with lower diet quality, and 4) non-smokers with higher diet quality.The hazard ratios (HRs) and their associated 95% confidence intervals (CIs) for each outcome were determined across the categories considering "current smokers with poor diet quality" as a reference group.Variables that had significant associations with CVDs and HTN according to the univariate analysis (P < 0.05), were included in the statistical analyses as confounding factors.Model 1 was adjusted for sex, age, SBP, FBS, serum TC, serum TG, serum HDL-C, energy intake, BMI, and physical activity.Additionally, Model 2 was adjusted for medication use, employment status, education level, and family history of premature CVDs.The fitness of the multivariable models was determined using Harrel's C index 48 .To obtain P values for interactions, smoking status, diet quality status, and their interaction terms were included in the unadjusted Cox model.The results include HR (95% CI) for each comparison.The proportional hazards assumption was checked using the graph of the log(−log(survival)) versus the log of survival time, and the assumption was satisfied (lines in the plots were parallel).All P values were derived from two-sided tests, and values less than 0.05 were considered to indicate statistical significance.

Figure 1 .
Figure 1.The hazard ratio and 95% confidence interval (CI) for the incidence of hypertension, all-cause mortality, stroke and cardiovascular diseases (CVDs) among current smokers vs. non-smokers (reference group): Tehran Lipid and Glucose Study.

Figure 2 .
Figure 2. Diagram of the study participants and follow-up.

Table 3 .
Hazard ratio (95% confidence intervals) for cardiovascular events, all-cause mortality, and hypertension based on the combined effects of diet quality and smoking status.DQI-I diet quality indexinternational, DQI-R diet quality index-revised.Model 1, was adjusted for sex, age, systolic blood pressure, fasting blood sugar, serum cholesterol, serum triglyceride, high-density lipoprotein cholesterol, energy intake, BMI, and physical activity.Model 2 additionally adjusted for medication use, employment status, education level, and family history of premature CVD.All diet quality indices were constructed according to each 1000 kcal of energy intake.