Cardiometabolic Profile in Young Adults With Diverse Cigarette Smoking Histories: A Longitudinal Study From Adolescence

Abstract

Background

For the first time, the present study investigated smoking trajectory and cardiometabolic profile from adolescence to young adulthood in a middle‐income developing country facing a high prevalence of smoking and cardiovascular disease‐related outcomes.

Methods and Results

Data on 1082 adolescents (12–18 years of age) who participated in the TLGS (Tehran Lipid and Glucose Study) were gathered, and participants were followed for a median of 12.5 years (baseline: 1999–2002, last follow‐up: 2014–2017). Participants were categorized as non/rare smokers, experimenters, and escalators using group‐based trajectory models. Statistical analysis was used to compare the trajectory groups' cardiometabolic components, clinical characteristics, and cardiometabolic changes due to the individuals' placement in experimenter and escalator groups compared with non/rare smokers. The smoking trajectory groups in young adulthood differ significantly in blood pressure, triglycerides, high‐density lipoprotein cholesterol, waist circumference, and body mass index, with the escalator group having the highest risk values for each component. Significant differences were observed in blood pressure (P=0.014), triglycerides (P<0.001), and waist circumference (P<0.001) status after using clinical cut points. The adjusted linear regression revealed that the escalator group had 3.16 mm Hg‐lower systolic blood pressure SBP (P=0.016), 2.69 mm Hg‐lower diastolic blood pressure (P=0.011), and 4.42 mg/dL‐lower high‐density lipoprotein cholesterol (P=0.002), compared with the non/rare smoker group.

Conclusions

Despite elevated risks in unadjusted analyses for all cardiometabolic components among smokers, our study identified a modest protective link between early smoking and blood pressure in addition to a remarkable harmful association with high‐density lipoprotein cholesterol levels exclusively in the escalator group during the developmental stage to young adulthood, using adjusted analyses.

Nonstandard Abbreviations and Acronyms

DBP

diastolic blood pressure

FBS

fasting blood sugar

PA

physical activity

SBP

systolic blood pressure

T‐Chol

total cholesterol

TLGS

Tehran Lipid and Glucose Study

WC

waist circumference

Clinical Perspective.

What Is New?

• Adolescents with varying smoking habits in a low/middle‐income population display divergent long‐term trends in cardiometabolic factors.

• Youth in the escalator smoking group experience a notable reduction in high‐density lipoprotein cholesterol levels.

• Our research also identifies an association between early‐onset smoking and reduced blood pressure in young adulthood.

What Are the Clinical Implications?

• Targeted smoking cessation interventions may significantly reduce cardiovascular risk factors such as blood pressure, triglyceride levels, and waist circumference in the escalator group, which has the highest risk values for cardiovascular risk factors.

• Identifying individuals in young adulthood who belong to the escalator group is crucial for implementing timely interventions and monitoring their cardiovascular risk factors, including blood pressure, triglyceride levels, and waist circumference to prevent or manage potential health complications.

• A comprehensive risk assessment incorporating smoking status and multiple cardiovascular risk factors, including blood pressure, triglyceride levels, high‐density lipoprotein cholesterol, waist circumference, and body mass index is essential for identifying individuals at higher risk and tailoring interventions to improve patient outcomes.

The cardiometabolic components represent metabolic abnormalities that precipitate cardiovascular disease (CVD) incidence, a globally predominant cause of death. ¹ Data spanning from 1990 to 2015 reveal a 27.3% decline in age‐standardized CVD mortality rates; however, deaths rose by 42.4%. ² According to the World Health Organization, >75% of CVD‐related deaths occur in low/middle‐income countries, ¹ imposing a strain on health care systems. In Iran, a low/middle‐income country, CVD is the primary cause of mortality, contributing to 50% of deaths and 20% to 23% of the CVD‐related disease burden, according to Global Burden of Disease's reports in 2010 and 2015. ³ Compared with 2005, projections for 2025 anticipate a 2‐fold increase in disability‐adjusted life‐years related to CVD in Iran, with a narrowing gender gap. ⁴

Smoking, the most prevalent form of tobacco use, emerges as a paramount global public health concern, ranking as the fifth actual cause of CVD‐related deaths. ⁵ Notably, in 2021, at least 1 in 5 men and 1 in 10 women who smoke are projected to eventually die from CVD. ⁶ Similar outcomes within the entire population of Iran have further highlighted the role of smoking, attributing 7.7% of all‐cause mortality to this behavior. ⁷ Studies have demonstrated that health‐related problems in adulthood are primarily rooted in the early years of life. ⁸ The link between initiating smoking in adolescence and establishing persistent smoking habits in adulthood is often neglected. ⁹ Global youth tobacco surveys spanning from 1990 to 2018 revealed that 11.3% of boys and 6.1% of girls in adolescence globally had smoked at least once in the past month. ¹⁰ Research conducted between 2000 and 2019 indicates an estimated 12% smoking prevalence in Iranian adolescent boys and 6% in girls. ¹¹

The complex period of adolescence and the intricate behavior of smoking could yield diverse health‐related statuses in young adulthood. Using trajectory modeling as a practical tool, recognized for capturing intra‐ and interindividual variability, proves invaluable in navigating the dynamic and intricate nature of smoking, influenced by multiple extrinsic factors. ¹² Categorizing participants into behaviorally similar subgroups in this approach enhances the model's efficacy in probing variable relationships compared with participant averages. ¹³ To the best of our knowledge, most studies of smoking trajectory groups have been conducted to investigate the factors responsible for membership in each trajectory group, ¹⁴ , ¹⁵ , ¹⁶ and such studies for investigating different aspects of health in adulthood study end points are limited. In regard to cardiometabolic consequences, only 1 study has investigated the correlation between smoking trajectory and cardiometabolic profile in a sample of 2009 Black individuals from adolescence toward adulthood, indicating that the early‐onset, light smoking class had the greatest cardiometabolic risk. ¹⁷

Most studies to identify smoking trajectories and their influential factors have been conducted in developed countries, and due to multifaceted differences, their results cannot be generalized to developing countries. As one of the first efforts in Iran, ¹⁸ our previous research in the TLGS (Tehran Lipid and Glucose Study), a family‐based longitudinal study conducted on a large Middle Eastern population, revealed smoking developmental trajectories from adolescence to young adulthood. By using a smoking score incorporating frequency and intensity, we discovered that adolescents go through 1 of non/rare smoker, experimenter, and escalator trajectory groups until young adulthood, and certain factors, including being of male sex, having a lower education level, being employed after 18 years of age, and having a father who smokes, can increase the likelihood of experimenting or continuing smoking with greater intensity. In the current study, we aimed to provide a detailed investigation of smoking trajectories in adolescence and their relationship with individuals' cardiometabolic profile in adulthood among a middle‐income developing country facing a high prevalence of smoking and CVD‐related outcomes. The results of this study provide data for targeted preventive planning tailored to each subgroup considering existing sociocultural differences with Western countries.

Methods

The data sets used during the current study are available from the corresponding author upon reasonable request.

This study was approved by the Ethical Committee of Research Institute for Endocrine Sciences and the National Research Council of the Islamic Republic of Iran (ethic number: 121). Informed consent was obtained from all individual participants included in the study. All procedures were in accordance with the ethical standards of the institutional or national research committee and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards.

Sampling and Participants

TLGS started in 1999 by selecting 3 health care centers through random cluster sampling among 20 health care centers located in the 13th District of Tehran. This district was chosen as a study sample due to population stability, full access to family documented data, similar age distribution, and socioeconomic status to the main population to identify risk factors related to noncommunicable diseases in a large‐scale cohort study with long‐term follow‐up examinations. After completing the consent forms, 7151 families, including 15 005 individuals (≥3 years of age), participated in the TLGS baseline assessment. Since 2002, to investigate changes in disease prevalence and related risk factors, data were collected from baseline participants cross‐sectionally and every 3 years in 5 follow‐up assessments (baseline assessment: 1999 to 2002, first follow‐up: 2002 until 2005, second follow‐up: 2005 to 2008, third follow‐up: 2008 to 2011, fourth follow‐up: 2011 to 2014, and fifth follow‐up: 2014 to 2017). More detailed information about the TLGS protocol, design, and data collection are discussed in other studies. ¹⁹ , ²⁰

For the current study, the sample is restricted to 1567 adolescents (12–18 years of age) who were followed up for a median of 12.5 years. After excluding 396 individuals with at least 3 missing data points on the smoking variable, 87 individuals with missing data on cardiometabolic components in young adulthood, and 2 individuals who died during follow‐up assessments, the analysis was performed using the data of 1082 participants.

Measurements

Demographic Data

In each follow‐up assessment, trained interviewers used valid and reliable questionnaires that included sociodemographic data, health status, and behavioral data. Education during adolescence was classified as illiterate/primary (0–6 years) and secondary (6–12 years) based on years of education, but in adulthood, education was divided into secondary or lower than secondary (0–12 years) and higher (>12 years). Considering marital status, the participants were divided into married and single groups only in adulthood. The participants in adolescence and adulthood were classified into 2 categories, employed and unemployed, based on having a job or not.

Health Data

The Persian translation of the Modified Activity Questionnaire was used to evaluate the physical activity (PA) of adolescents and adults. ²¹ , ²² Participants reported their PA type (including leisure and occupational activities) as well as the frequency and duration of each activity in the past 12 months. The number of minutes per week for each activity was multiplied by its metabolic equivalent. After that, total PA was calculated, and its levels were defined as low (<600 minutes per week) and moderate/high (≥600 minutes per week).

A digital scale (Seca 707: range 0–150 kg with an accuracy of 100 g) and a tape stadiometer were used to measure weight and height with standard protocols. The participants' shoes and heavy clothes were removed for accurate weight measurement, and the participants' head, shoulders, hips, and heels were attached to the wall for accurate height measurement. Body mass index (BMI) was calculated by dividing body weight by the square of height in kilograms per square meter. Participants were given a 15‐minute break to be prepared for blood pressure (BP) measurement. BP was measured twice with an interval of at least 30 seconds using a standard mercury sphygmomanometer calibrated to the participant's arm. After inflating the cuff to 30 mm Hg above the pressure at which the radial pulse was not palpable, the cuff was deflated at a rate of 2 to 3 mm Hg/s. The first and fifth Korotkoff sounds were defined as systolic and diastolic blood pressure (SBP and DBP), respectively. The average BP measured on 2 experiences was reported. The same personnel recorded all measurements to eliminate subjective errors.

Laboratory Data

After a 12‐ to 14‐hour overnight fasting, participants' blood samples were collected between 7:00 and 9:00 am in Vacutainer tubes and centrifuged within 30 to 45 minutes of collection. Fasting blood sugar (FBS) was measured by the enzymatic colorimetric glucose oxidase method. Two‐hour postprandial blood glucose was measured 2 hours after receiving 75 g of oral glucose using an enzymatic calorimetry method with the glucose oxidase technique. In the TLGS research laboratory, on the blood collection day, a Selectra 2 autoanalyzer (Vital Scientific, Spankeren, the Netherlands) was used to analyze the samples for serum total cholesterol (T‐Chol) and triglycerides. Enzyme colorimetric tests were used to measure T‐Chol with cholesterol esterase and cholesterol oxidase. For triglycerides, glycerol phosphate oxidase was used. High‐density lipoprotein cholesterol (HDL‐C) was measured after the precipitation of lipoproteins containing lipoprotein B with phosphotungstic acid. Friedewald formula was used to calculate low‐density lipoprotein cholesterol in cases where triglycerides were <400 mg/dL for adolescents and adults. ²³ , ²⁴ Both inter‐ and intra‐assay coefficients of variation for laboratory measurements were insignificant, and all samples were analyzed when the internal quality control met the acceptable criteria.

Definitions of Terms

General obesity was determined as normal (18.5 ≤ BMI < 25 kg/m²), overweight (25 ≤ BMI < 30 kg/m²), and obese (BMI ≥30 kg/m²). In regard to Adult Treatment Panel III and the joint interim statement, ²⁵ abnormal cardiometabolic components were classified as: (1) triglyceride levels ≥150 mg/dL or receiving medications for elevated triglycerides; (2) T‐Chol levels ≥200 mg/dL or receiving medications for elevated T‐Chol; (3) HDL‐C <50 mg/dL in female participants, <40 in male participants, or receiving medications for reduced HDL‐C; (4) low‐density lipoprotein cholesterol levels ≥130 mg/dL or receiving medications for elevated low‐density lipoprotein cholesterol; (5) elevated BP (≥130 mm Hg SBP or ≥ 85 mm Hg DBP) or receiving antihypertensive medications; (6) elevated FBS ≥100 mg/dL or receiving medications for elevated FBS; and (7) waist circumference (WC) ≥90 cm for both sexes.

Trajectory Variable

At each follow‐up measurement, smoking data were obtained using standardized questionnaires. Adolescents (≤18 years of age) reported their smoking quantity (number of cigarettes smoked per day) and smoking frequency (number of smoking days in the past 30 days). The smoking quantity was categorized into <1, 1 to 5, 6 to 10, 11 to 19, and ≥20 (1–5 score), and the smoking frequency was classified as 0, 1 to 2, 3 to 14, 15 to 29, an 30 (0–4 score). Then, categorical quantity and frequency variables were multiplied to obtain a single outcome measure (smoking score) that ranged from 0 to 20.

Adults (>18 years of age) answered a question about their current smoking with yes/every day, yes/sometimes, or no. If the answer to the first question was positive, the number of cigarettes consumed per day (quantity) and the number of smoking days in the past 7 days (frequency) were assessed. The researchers converted the number of smoking days in the past 4 days to those in the past 30 days. Then, like adolescents, each variable was classified and then multiplied to create the smoking score in adults.

Statistical Analysis

To perform trajectory analysis, adolescents who had 3 out of 5 smoking variables during the study assessments were included in the study. This study used group‐based trajectory models to identify similar subgroups of participants who shared similar underlying changes in the smoking index from 12 to 32 years of age. The analytic sample was 12‐ to 18‐year‐old participants of baseline who were 28 to 32 years of age in the last follow‐up. Due to the high prevalence of never‐smokers (0 values for smoking index) among adolescents, 0 inflated Poisson trajectory models with a user‐written program Traj (Data S1) were estimated. ²⁶ The 0‐inflated model is used for analyzing count data with excess 0 counts (0‐inflated in the smoking score). We investigated the possible number of latent trajectories by a series of models considering several linear and polynomial (cubic, quadratic) specifications of the smoking score as a function of age (centralized at 20 years of age). In this model, the intercept represents the initial smoking score for the 20‐year‐old individuals at baseline. The linear or nonlinear changes in smoking score according to age were determined using linear, quadratic, or cubic slopes for each trajectory group. We began with a single model consisting of 1 group and then increased the number of groups until we determined the best fit of group numbers for trajectories. To obtain the number of trajectories that best fit the data, the Bayesian information criterion, the average posterior probability of group membership, and the odds of correct classification were calculated for 1 to 4 trajectories. The optimal model was selected as the lowest Bayesian information criterion, highest average posterior probability >0.70, odds of correct classification >5, and the minimum of 5% of the total sample for the size of each class. The 3‐group model with a linear function of age for the first group and cubic function of age for the second and third groups was selected with Bayesian information criterion=–2509.77; average posterior probability=0.95, 0.93, 0.94; and odds of correct classification=5.6, 88.6, 154.1 (model number 12 in Table S1), respectively, for group membership 1 to 3. Based on the mentioned content, the study participants were divided into 3 groups, including non/rare smokers (n=842), experimenters (n=142), and escalators (n=98). More detailed information on smoking trajectories has been discussed in another article. ¹⁸

Participants' characteristics were compared among smoking trajectory groups using the χ² test and 1‐way ANOVA for categorical and continuous variables, respectively. The association of the cardiometabolic profile at adulthood in the last follow‐up examination with the smoking trajectories was assessed using the linear regression analysis. The regression coefficients and 95% CIs were estimated. The regression coefficients represent the difference in the value of each cardiometabolic component for trajectory groups compared with the non/rare smokers. The models were adjusted for age, sex, education level, occupation, marital status, physical activity, and BMI at the last follow‐up examination. The changes in cardiometabolic components from adolescence to young adulthood were plotted and assessed using the generalized estimating equations model. The trajectory analysis was conducted using Stata software version 16, and the remaining analysis was conducted in IBM SPSS Statistics version 26. Two‐sided P values <0.05 were considered statistically significant.

Results

Data on 1082 adolescents (45.6% boys) who participated in the TLGS from 1999 to 2002 were gathered, and participants were recruited in the current analysis and followed for a median of 12.5 years (last follow‐up: 2014–2017). Table S2 represents participants' sociodemographic and cardiometabolic characteristics in adolescence (baseline) and early adulthood (last follow‐up examination). The mean baseline age of adolescents was 15.18±1.97 years. Although 43.1% of the participants had a low PA in adolescence, this rate increased to 54.2% in adulthood. Except for triglycerides, other cardiovascular components showed higher values at the last follow‐up examination compared with the baseline.

Adulthood sociodemographic and cardiometabolic components (at the last follow‐up examination) in each smoking trajectory group are presented in Table 1. Male participants comprised 36.1% of non/rare smokers, and they formed the majority in the experimenters (70.4%) and escalators (90.8%), which differed significantly among groups. The escalator group had the highest proportion of single, less‐educated, and employed participants (P=0.004, P<0.001, and P<0.001, respectively) compared with the other groups. The majority of participants in all 3 groups had low PA in adulthood; however, there was no significant difference among them. An unadjusted comparison of the cardiometabolic profile in adulthood among the 3 smoking trajectories showed significant differences, so that SBP, DBP, triglycerides, HDL‐C, WC, and BMI were significantly different among the 3 trajectories, with the highest values in the escalator group for each component. However, no significant difference was observed among groups on T‐Chol and low‐density lipoprotein cholesterol in lipid profile and glucose‐related factors like FBS and 2‐hour postprandial blood glucose. The Figure shows the changes in cardiometabolic components from adolescence to young adulthood among the 3 groups. Significant increasing trends were observed in SBP, DBP, triglycerides, HDL‐C, WC, and BMI among the smoking trajectory groups over the study assessments (all P values <0.05).

Table 1.

Adult Participants' Characteristics in the Last Follow‐Up Examination According to Smoking Trajectory Group

Characteristic	Non/rare smokers, n=842	Experimenters, n=142	Escalators, n=98	P value
Age, y	27.54 (2.70)	27.92 (2.46)	28.18 (2.70)	0.032
Women, %	63.9% (538)	29.6% (42)	9.2% (9)	<0.001
Education level, %				<0.001
Diploma or lower	29.6% (249)	31.7% (45)	51.0% (50)
Higher than diploma	70.4% (593)	68.3% (97)	49.0% (48)
Occupation, %				<0.001
Unemployed	47.4% (399)	23.2% (33)	12.2% (12)
Employed	52.6% (442)	76.8% (109)	87.8% (86)
Marital status, %				0.004
Married	51.0% (428)	39.4% (56)	37.8% (37)
Single	49.0% (412)	60.6% (86)	62.2% (61)
Physical activity, %				0.800
Low	54.8% (311)	53.6% (59)	50.7% (37)
Moderate/high	45.2% (257)	46.4% (51)	49.3% (36)
SBP, mm Hg	105.24±11.92	108.72±13.59	108.92±10.56	<0.001
DBP, mm Hg	71.43±8.60	73.80±8.97	73.29±8.76	0.003
WC, cm	85.76±11.75	90.08±13.23	93.43±12.82	<0.001
Triglycerides, mg/dL	89.00 (66–128)	99.00 (72–170)	113.00 (85–152)	<0.001
T‐Chol, mg/dL	173.07±34.85	173.52±38.15	174.29±38.03	0.949
HDL‐C, mg/dL	50.30±11.48	46.14±10.83	42.04±9.31	<0.001
LDL‐C, mg/dL	99.61±28.35	101.90±30.36	105.46±31.28	0.433
FBS, mg/dL	88.45±9.53	88.96±9.26	90.18±14.02	0.258
2hpp, mg/dL	97.51±21.77	97.14±25.90	93.99±22.86	0.406
BMI, kg/m2	25.32±4.60	26.33±5.56	26.97±4.72	0.001

Data are presented as percentage (number) or mean±SD. Data for triglycerides are presented as median (Q1–Q3). The P value is for comparison among smoking trajectory groups, achieved by conducting a χ² test and 1‐way ANOVA for categorical and continuous variables, respectively. 2hpp indicates 2‐hour postprandial blood glucose; BMI, body mass index; DBP, diastolic blood pressure; FBS, fasting blood sugar; HDL‐C, high‐density lipoprotein cholesterol; LDL‐C, low‐density lipoprotein cholesterol; SBP, systolic blood pressure; T‐Chol, total cholesterol; and WC, waist circumference.

Figure . The cardiometabolic changes in non/rare smokers, experimenters, and escalators from baseline to the fourth follow‐up examinations over a median of 12.5 years.

These changes in SBP, DBP, TG, HDL, WC, and BMI were significant among the 3 groups using generalized estimating equations analysis. 2hpp indicates 2‐hour postprandial blood glucose; BMI, body mass index; CHOL, total cholesterol; DBP, diastolic blood pressure; FBS, fasting blood sugar; HDL‐C, high‐density lipoprotein cholesterol; LDL‐C, low‐density lipoprotein cholesterol; SBP, systolic blood pressure; TG, triglycerides; and WC, waist circumference.

Table 2 displays the clinical characteristics of participants in each trajectory group, using clinical cut points for BP, FBS, lipid profile, BMI, and WC. Smoking trajectory groups had significant clinical differences in terms of BP (P=0.014), triglycerides (P<0.001), and WC (P<0.001) status. The escalator group had the most cases with elevated WC (56.7%) and included 63.9% of overweight and obese individuals. On the other hand, the experimenters had a higher proportion of individuals with elevated BP and triglycerides, comprising 14.9% and 35.3%, respectively.

Table 2.

Adult Participants' Statuses in the Last Follow‐Up Examination According to Smoking Trajectory Group

Status	Non/rare smokers	Experimenters	Escalators	P value
BP status				0.014 *
Normal	92.2% (765)	85.1% (120)	87.6% (85)
Elevated BP	7.8% (65)	14.9% (21)	12.4% (12)
Triglyceride status				<0.001 *
Normal	82.3% (682)	64.7% (90)	72.2% (70)
Elevated triglycerides	17.7% (147)	35.3% (49)	27.8% (27)
Cholesterol status				0.395
Normal	80.3% (666)	76.3% (106)	76.3% (74)
Elevated T‐Chol	19.7% (163)	23.7% (33)	23.7% (23)
HDL‐C status				0.686
Normal	63.4% (526)	59.7% (83)	61.9% (60)
Reduced HDL‐C	36.6% (303)	40.3% (56)	38.1% (37)
LDL‐C status				0.150
Normal	86.5% (717)	84.2% (117)	79.4% (77)
Elevated LDL‐C	13.5% (112)	15.8% (22)	20.6% (20)
Blood sugar status				0.484
Normal	91.3% (757)	94.2% (131)	90.7% (88)
Elevated FBS	8.7% (72)	5.8% (8)	9.3% (9)
BMI status				0.082
Normal	50.4% (411)	44.6% (62)	36.1% (35)
Overweight	35.0% (285)	38.8% (54)	43.3% (42)
Obesity	14.6% (119)	16.5% (23)	20.6% (20)
WC status				<0.001 *
Normal	66.2% (540)	56.1% (78)	43.3% (42)
Higher WC	33.8% (276)	43.9% (61)	56.7% (55)

Data are presented as percentage (number). The P value is for comparison among smoking trajectory groups, achieved by conducting a χ² test. BP indicates blood pressure; BMI, body mass index; FBS, fasting blood sugar; HDL‐C, high‐density lipoprotein cholesterol; LDL‐C, low‐density lipoprotein cholesterol; T‐Chol, total cholesterol; and WC, waist circumference.

^*Indicates significant values.

After controlling for the participant's age, sex, education level, occupation, marital status, physical activity, and BMI, the association of adulthood cardiometabolic components with the smoking trajectories is presented in Table 3. The significant association between early smoking and cardiometabolic components was observed only among escalators and in SBP (P=0.016), DBP (P=0.011), and HDL‐C (P=0.002). On average, the escalator group had lower SBP (b=−3.16 mm Hg [95% CI, −5.74 to −0.58]), lower DBP (b=−2.69 mm Hg [95% CI, −4.75 to −0.63]), and lower HDL‐C (b=−4.42 mg/dL [95% CI, −7.14 to −1.70]), compared with the non/rare smoker group. Although there was no significant difference among study groups on other cardiometabolic components, the difference in mean values (compared with non/rare smokers) in escalators was greater than the experimenters, except for FBS.

Table 3.

Association of the Cardiometabolic Components With the Smoking Trajectories Among Adults in the Last Follow‐Up Examination: Linear Regression Results

Variable		SBP (mm Hg)	DBP (mm Hg)	WC (cm)*	Triglycerides (mg/dL)	T‐Chol (mg/dL)	HDL‐C (mg/dL)	LDL‐C (mg/dL)	FBS (mg/dL)	2hpp (mg/dL)	BMI (kg/m2)*
Non/rare smokers	b	Ref	Ref	Ref	Ref	Ref	Ref	Ref	Ref	Ref	Ref
	95% CI
	P
Experimenters	b	−0.27	−0.80	0.83	2.71	−3.92	−0.64	−3.23	−0.97	−2.72	0.09
	95% CI	−2.36 to 1.83	−2.47 to 0.88	−1.44 to 3.10	−11.82 to 17.23	−11.47 to 3.64	−2.86 to 1.59	−9.20 to 2.75	−2.84 to 0.89	−7.54 to 2.10	−0.089 to 1.07
	P	0.804	0.351	0.474	0.714	0.309	0.574	0.289	0.305	0.267	0.852
Escalators	b	−3.16	−2.69	1.05	4.72	−8.58	−4.42	−4.22	−0.12	−3.25	0.16
	95% CI	−5.74 to –0.58	−4.75 to –0.63	−1.75 to 3.84	−13.09 to 22.52	−17.85 to 0.68	−7.14 to –1.70	−11.54 to 3.10	−2.41 to 2.17	−9.33 to 2.82	−1.05 to 1.37
	P	0.016†	0.011†	0.463	0.603	0.069	0.002†	0.258	0.920	0.293	0.796

The coefficient represents the difference in the value of each cardiometabolic component for trajectory groups compared with non/rare smokers over study assessments. The models were adjusted for age, sex, education level, occupation, marital status, physical activity, BMI, and the last follow‐up examination. 2hpp indicates 2‐hour postprandial blood glucose; BMI, body mass index; DBP, diastolic blood pressure; FBS, fasting blood sugar; HDL‐C, high‐density lipoprotein cholesterol; LDL‐C, low‐density lipoprotein cholesterol; Ref, reference; SBP, systolic blood pressure; T‐Chol, total cholesterol, and WC, waist circumference.

^*For BMI and WC as a component, BMI was not involved in adjustment.

^† Indicates significant values (P<0.05).

Discussion

This study was conducted to investigate the association between diverse cigarette smoking histories from adolescence and cardiometabolic profile in young adulthood through >12.5 years of follow‐up (baseline: 1999–2002, last follow‐up: 2014–2017). Participants were classified into 3 main trajectories based on their smoking patterns from their adolescence, including non/rare smokers, experimenters, and escalators. Although our unadjusted comparisons generally revealed higher risk in smokers, mainly in the escalators, this result was clinically and statistically significant only in BP, triglycerides, and WC. However, after controlling for potential young adulthood confounders, a slight protective association between early smoking and BP, along with a remarkable harmful relationship with HDL‐C levels, have been observed only in the escalator group during this developmental stage to young adulthood. A similar result was not observed in experimenters using this data context.

Considering participant characteristics, as discussed in our previous study, ¹⁸ non/rare smokers, comprising the largest group, maintained a consistent 0 smoking score throughout the study assessments. The experimental group exhibited an increasing trend in smoking score, commencing at 16 years of age, followed by a stable smoking score from 24 to 30 years of age, and a further increasing trend thereafter. In contrast, participants in the escalator group initiated an increase in smoking scores at a younger age than the experimental group and consistently increased their smoking frequency or quantity throughout the study.

In this study, we investigated young adulthood cardiometabolic components as an outcome of smoking developmental trajectories. An overview of the cardiometabolic component values for the mentioned groups revealed that non/rare smokers had the most favorable status in relation to all cardiometabolic components, except for 2‐hour postprandial blood glucose. However, the escalators were opposite to non/rare smokers in all cardiometabolic components. The association of smoking with some factors affecting the individuals' cardiometabolic profile, as well as more adherence to other unhealthy behaviors along with smoking, makes it possible to expect such results. ²⁷ , ²⁸ On the other hand, in the unadjusted analysis, we observed that although the smoking trajectories differed in young adulthood cardiometabolic components such as SBP, DBP, WC, BMI, triglycerides, and HDL‐C, this difference occurred mainly within the normal ranges. Significant differences were observed among trajectories, particularly in BP, WC, and triglyceride levels, after using clinical cut points. Because the association between smoking and individuals' cardiometabolic profile has been proven in various studies, ²⁹ our results can be justified based on the evidence of the lowest prevalence of cardiometabolic disorders in young adults compared with older age groups in Iran and other countries. ³⁰ , ³¹ , ³² , ³³ The possible reason might be that the observed timeframe was not long enough to detect a significant impact of smoking on the participants' health. On the other hand, smoking participants were in ascending status in the smoking pattern at the end of the study, which suggests the hypothesis that the cumulative effect of higher levels of smoking will create more significant cardiometabolic results at older ages.

In our study, data analysis with the adjusted model showed that smoking had caused a significant difference in the SBP and DBP reduction (b=−3.16 mm Hg and b=−2.69 mm Hg, respectively) as well as HDL‐C (b = −4.42 mg/dL) in escalators compared with non/rare smokers. Our results are in line with various studies in Japan, ³⁴ , ³⁵ England, ³⁶ China, ³⁷ , ³⁸ United States, ³⁹ , ⁴⁰ Sweden, ⁴¹ Nepal, ⁴² and a meta‐analysis of 23 population‐based studies. ⁴³ Although the lack of belief and adoption of preventive health behaviors and, as a result, lower cardiovascular health in smokers, has been more remarkable, ²⁷ , ⁴⁴ this result can occur due to the study population, the analysis model, and unmeasured confounders (such as lifestyle, diet, and alcohol consumption). Additionally, the body's attempt to adapt to smoking through vascular angiogenesis, which lowers BP, can also be the cause of such a phenomenon. ³⁴ , ⁴⁵ On the other hand, smoking can alter the critical enzymes of the lipid transport cycle and lead to a reduction in HDL‐C quantity and function. ⁴⁶ Our result is aligned with another study that showed that smokers had lower adjusted levels of HDL‐C than nonsmokers. ⁴⁷

Conducting studies that delve into the clinical attributes of these trajectories, in conjunction with existing research exploring smoking patterns and the distinctive features of individuals within each trajectory, can furnish health policymakers with vital insights to solve public health issues. Conducting additional studies within the context of smoking trajectory groups holds promise for yielding valuable insights for implementing preventive interventions. For instance, a study has demonstrated that the outcomes of preventive policies are not uniform across all categories of smokers. ⁴⁸ This study indicates that the implementation of smoke‐free laws is correlated with diminished smoking initiation and reduced usage frequency across all trajectory groups, excluding experimenters, whereas higher tax rates demonstrate an association with decreased smoking initiation and frequency of use for all trajectories, except for escalators. These findings offer valuable models for formulating preventive policies in diverse national contexts.

Most smoking trajectory studies have been conducted in developed countries, and their results cannot be generalized to the developing countries of the Middle East region. This study, for the first time, investigated smoking trajectory and cardiometabolic profile from adolescence to young adulthood in a middle‐income developing country facing a high prevalence of smoking and CVD‐related outcomes using TLGS, as one of the oldest and most reliable cohorts in this region with a long‐term follow‐up. However, this study also had limitations. Due to the group‐based trajectory model nature (in which the members of each subgroup should constitute >5% of the participants' population) as well as the fact that smoking is considered taboo among women in the study population (which might lead to underreporting), it was not possible to conduct a gender‐specific study. In addition, the higher number of participants in the smoking groups will provide clearer results of the studies. Moreover, the community‐based nature of the TLGS did not make it possible to check the concentration of blood and urine biomarkers as a more accurate indicator for smoking to assess its long‐term association with cardiometabolic components. Smoking data were collected via questionnaires, which could increase the possibility of underreporting and misclassification. Additionally, data related to nutrition or high‐risk behaviors that are known to co‐occur with smoking, such as drug and alcohol use, which also have an impact on cardiometabolic components, were not available and thus not included in the current study. Conducting this study in a larger population that includes more individuals in the smoking groups and tracks gender differences over a more extended period into adulthood would help clarify these results.

Conclusions

The present study investigated smoking trajectory and cardiometabolic profile from adolescence to young adulthood (from baseline assessments in 1999–2002, to the last follow‐up assessments in 2014–2017) in a middle‐income developing country facing a high prevalence of smoking and CVD‐related outcomes. Although the engagement in smoking during adolescence exhibited an elevated risk across all cardiometabolic components, the magnitude of alterations was such that clinically significant differences were found solely in BP, triglycerides, and WC. In adjusted analyses, smoking was associated with an increased risk for HDL‐C and a mild protective relationship for BP only in the escalator group. Considering the study's limitations and observing the higher risk of elevated BP in smokers compared with non/rare smokers in unadjusted analyses and the mild protective relationship in adjusted analyses, it can be hypothesized that smoking might affect BP through other health variables.

Sources of Funding

None.

Disclosures

None.

Supporting information

Data S1 Supplemental Methods, Tables S1–S2.