Identifying smoker subgroups with low readiness to quit: a cluster analysis of perceived lung cancer and asthma risk profiles

Abstract

Background

Perceived disease risk is a key determinant of smoking cessation, yet many smokers underestimate their personal vulnerability to smoking-related diseases. This study identified distinct risk-perception profiles for lung cancer and asthma among smokers and examined their associations with sociodemographic factors, nicotine dependence, and readiness to quit.

Methods

A cross-sectional study was conducted among 310 adult cigarette smokers in Khuzestan Province, Iran (January–September 2024). Perceived susceptibility and perceived severity of lung cancer and asthma, nicotine dependence, and stage of change (readiness to quit) were assessed using validated instruments. Hierarchical cluster analysis (Ward’s method, Euclidean distance) was used to identify risk-perception profiles. Associations between clusters and participant characteristics were examined using Chi-square and Kruskal–Wallis tests. Multinomial logistic regression was fitted to identify independent predictors of cluster membership.

Results

Four distinct risk-perception profiles emerged for both diseases: Indifferent (low susceptibility/low severity), Optimistic Bias (low susceptibility/high severity), Dismissive (high susceptibility/low severity), and Concerned (high susceptibility/high severity). In adjusted analyses, the Indifferent cluster was independently characterized by low nicotine dependence, whereas the Concerned cluster combined high dependence with greater readiness to quit. In the asthma model, the Optimistic Bias cluster—despite low perceived personal susceptibility—displayed the most favorable quit readiness profile, including the highest proportion planning to quit within one month. Male gender independently predicted the Optimistic Bias cluster, and higher income reduced its likelihood in the asthma model; for lung cancer, low income and moderate or high nicotine dependence were significant independent predictors.

Conclusion

Smokers exhibit heterogeneous risk-perception profiles that are distinctly linked to nicotine dependence and quit readiness. The finding that the Indifferent profile is marked by low rather than high dependence suggests that lack of risk perception in this group stems from other cognitive or socioeconomic barriers beyond addiction. Tailored interventions should move beyond generic risk messages: for Indifferent smokers, addressing underlying barriers to risk acknowledgment; for Optimistic Bias smokers, leveraging their high perceived severity; and for Concerned smokers, coupling cessation support with dependence treatment.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12889-026-27876-2.

Introduction

Tobacco use remains one of the most pressing public health challenges worldwide, contributing to millions of deaths that are preventable each year [1]. Smoking is a major risk factor for numerous chronic diseases, including lung cancer and asthma, which together contribute significantly to global morbidity and mortality [2–4]. Despite widespread public health campaigns and the implementation of tobacco control policies, smoking prevalence continues to remain high in many low- and middle-income countries, such as Iran [5–7].

In Iran, tobacco use continues to impose a substantial health burden, with millions of active smokers and a considerable proportion of deaths attributable to smoking-related diseases. The prevalence of smoking among adults remains notable, and chronic respiratory conditions such as asthma further add to the national burden of disease [8, 9]. These data highlight the urgent need for more effective, theory-driven, and targeted smoking cessation interventions.

Effective prevention and cessation strategies depend on a clear understanding of how smokers perceive the health risks associated with tobacco use [10, 11]. In behavioral health models such as the Health Belief Model (HBM), perceived risk is a central construct that reflects an individual’s beliefs regarding their susceptibility to disease and the potential severity of its consequences [12]. Generally, individuals who perceive themselves to be at higher risk of tobacco-related diseases are more likely to attempt smoking cessation or engage in preventive health behaviors [13]. However, many smokers tend to underestimate their vulnerability to conditions such as lung cancer and asthma, which may diminish their motivation to quit [10, 14].

Risk perception in health behavior research has been operationalized using both absolute and comparative approaches.

This study focuses on comparative risk perception, where individuals evaluate their risk relative to an average smoker, as it more effectively captures cognitive biases such as optimistic bias and social comparison processes that are highly relevant in smoking behavior [15, 16], and may provide additional predictive value beyond absolute risk perceptions [17]. Absolute risk perception was not included because it requires individuals to estimate their objective probability of disease, which is often less intuitive and more prone to measurement error, particularly in populations with limited health numeracy.

Such underestimation and cognitive distancing can be attributed to several factors. Cultural norms and social influences often downplay the risks associated with smoking, leading to a collective minimization of perceived harm [10, 18, 19]. Optimistic bias, where individuals believe they are less likely to experience negative health outcomes compared to others, is prevalent among smokers and contributes to this underestimation [10]. Additionally, low health literacy and limited access to accurate health information can hinder smokers’ understanding of the risks associated with tobacco use, further perpetuating misconceptions about their susceptibility to diseases like lung cancer and asthma [10, 20].

In numerous studies, perceived risk has been quantified by calculating the sum or average of individual scores for its two components—perceived susceptibility and perceived severity [10, 21, 22]. Although widely used, this variable-centered approach assumes homogeneity in how these dimensions contribute to risk perception. However, different combinations of susceptibility and severity may yield identical total scores, thereby masking meaningful heterogeneity and limiting construct validity [23–26].

To address this limitation, recent methodological advances have emphasized person-centered approaches such as cluster analysis, which identify subgroups with distinct cognitive profiles that cannot be captured using traditional composite scoring methods [27–29]. This approach enables a more nuanced understanding of heterogeneity in risk perception and its association with smoking behavior and readiness to quit.

Despite growing interest in risk perception research, evidence remains limited regarding how combined patterns of perceived susceptibility and severity differ across multiple smoking-related diseases, particularly in Middle Eastern populations. Moreover, few studies have simultaneously applied person-centered approaches to both lung cancer and asthma risk perception profiles.

Cluster analysis is a widely used exploratory method in health behavior research, enabling the identification of subgroups based on patterns of perception or behavior, which can then be compared to determine associated risk factors [28, 30–33].

Building on this literature, the present study addresses gaps in prior research by simultaneously examining risk perception profiles for lung cancer and asthma among smokers in Khuzestan Province using a person-centered approach. Specifically, we employed cluster analysis to identify subgroups with distinct risk perception patterns based on perceived susceptibility and severity scores. Furthermore, we investigated how these profiles relate to a widely used measure of nicotine dependence(measured by the Fagerström Test), stages of readiness to quit smoking, and key sociodemographic characteristics. This study aimed to provide a comprehensive mapping of risk perception heterogeneity across two smoking-related diseases, which is essential for informing tailored smoking cessation interventions targeting distinct cognitive subgroups.

Methods

Procedure

To investigate the factors influencing smokers’ perceived risk of lung cancer and asthma, a cross-sectional study was conducted between January and September 2024. Eligibility criteria included adults aged 18 years or older who reported smoking cigarettes at least once daily. Individuals with a self-reported history of lung cancer or asthma were excluded.

Initially, 322 individuals were invited to participate in the study. Of these, 12 were excluded prior to analysis, including 8 due to a self-reported history of asthma or lung cancer and 4 due to incomplete questionnaires. Accordingly, the final analytical sample consisted of 310 participants.

Participants were recruited using a convenience sampling approach from public settings, including parks, markets, coffee houses, and workplaces, across six cities in Khuzestan Province, southwestern Iran.

Measures

Data were collected using a structured questionnaire comprising sections on sociodemographic and behavioral characteristics, perceived risk of lung cancer and asthma, stage of change (representing readiness to quit smoking), and attitudes and beliefs toward smoking. Demographic variables included age, gender, education level, occupation, duration and intensity of cigarette smoking, and previous quit attempts. To evaluate the perceived risk of cigarette smoking, questionnaires adapted from those designed by Wong and Cappella and Kotz et al. for lung cancer, and previously used in Iran by Zarghami et al., were employed with additional questions focusing on the perceived risk of asthma [10].

Nicotine dependence was measured using the Fagerström Test for Nicotine Dependence (FTND), a widely used instrument for assessing the intensity of physical addiction to nicotine [34]. The FTND consists of six items evaluating smoking behaviors such as time to first cigarette after waking, number of cigarettes smoked per day, and difficulty refraining from smoking in restricted areas, with total scores ranging from 0 to 10. For analysis, FTND scores were categorized into three levels of dependence based on commonly used cutoffs in the literature: low (0–4), moderate (5–7), and high (8–10) [35]. These categories were used to define the clustering groups for all subsequent analyses in this study.

A Persian version of the Fagerström Test for Nicotine Dependence (FTND) has been psychometrically evaluated among Iranian smokers, demonstrating acceptable validity and reliability for measuring nicotine dependence in this population. The study reported satisfactory internal consistency and stability, indicating that the Persian adaptation is suitable for use in research and tobacco control programs [36]. Additionally, this instrument has been used in previous research in Iran to assess nicotine dependence, including studies examining its association with perceived health risks among smokers [21].

The risk perception questionnaire comprised two sections: participants’ perceived risk of developing lung cancer and asthma, and their perceived likelihood of survival if they were to develop these conditions. The perceived risk section included two questions assessing participants’ beliefs about their likelihood of developing each disease compared with non-smokers of the same age. Responses were rated on an 11-point scale ranging from 0 (very unlikely) to 10 (very likely). Higher scores indicated greater perceived susceptibility to disease.

The second section assessed participants’ beliefs about their probability of surviving for five years if diagnosed with lung cancer or asthma. A score of 0 indicated that participants believed they were very likely to survive, whereas a score of 10 indicated that they believed survival was very unlikely. Thus, higher scores reflected greater perceived severity (lower survival).

The stage of change in cigarette smoking was measured using the framework developed by Velicer et al., based on the Transtheoretical Model of behavior change. This instrument classifies individuals into five stages—pre-contemplation, contemplation, preparation, action, and maintenance—according to their readiness to quit smoking. In this study, the stages were defined as follows: (1) pre-contemplation stage – currently smoking and not intending to quit within the next six months; (2) contemplation stage – considering quitting within the next six months; (3) preparation stage – planning to quit within the next month; (4) action stage – quit smoking but for less than six months; and (5) maintenance stage – quit smoking for more than six months [37]. The Persian version of the questionnaire, previously adapted and validated by Davaji et al., was used in this study [38].

Attitudes and beliefs about cigarette smoking were assessed using three Likert-scale items with response options ranging from “completely agree” to “completely disagree.” The items included the following statements: “Cigarette smoking is not addictive,” “I can stop smoking cigarettes whenever I decide,” and “The harms of cigarette smoking are less than those of hookah smoking.” Participants who disagreed with these statements were categorized as having a negative attitude toward cigarette smoking, whereas those who agreed were considered to have a positive attitude. This instrument was adapted from a questionnaire previously used by Davaji et al. [39]. For analysis, attitude scores were dichotomized into positive and negative categories. Full survey items available in Supplement 1.

Statistical analysis

Descriptive statistics were used to summarize participants’ sociodemographic characteristics, smoking behaviors, and perceived risk measures. Continuous variables are presented as mean and standard deviation (SD), while categorical variables are reported as frequencies and percentages.

Perceived susceptibility and perceived severity scores for lung cancer and asthma were categorized into three levels: low (scores < 4), moderate (scores 4–7), and high (scores > 7), based on predefined cutoffs.

To identify subgroups of smokers with similar risk-perception patterns, separate hierarchical cluster analyses were conducted for lung cancer and asthma using perceived susceptibility and perceived severity scores. Ward’s method with squared Euclidean distance was applied to derive homogeneous clusters. Based on the resulting patterns and mean scores within each cluster, four interpretable profiles were identified and labeled as Indifferent (low susceptibility/low severity), Optimistic Bias (low susceptibility/high severity), Dismissive (high susceptibility/low severity), and Concerned (moderate-to-high susceptibility and severity).

Nicotine dependence was assessed using the Fagerström Test for Nicotine Dependence (FTND) and categorized into three levels based on standard cutoffs: low (scores 0–4), moderate (scores 5–7), and high (scores 8–10). These categories were treated as a categorical variable in subsequent analyses.

Associations between cluster membership and categorical variables (e.g., gender, education, income level, additional tobacco use, stage of change, and nicotine dependence level) were examined using Pearson’s Chi-square tests. When overall associations were statistically significant, post hoc pairwise comparisons were conducted using Chi-square tests with Bonferroni adjustment to control for multiple comparisons.

For continuous variables (e.g., age, age at smoking initiation, and number of cigarettes smoked per day), differences across clusters were assessed using the Kruskal–Wallis test. When significant, post hoc pairwise comparisons were performed using Dunn’s test with Bonferroni correction.

Multinomial logistic regression was used to identify independent predictors of risk perception cluster membership. Separate models were fitted for asthma and lung cancer clusters, with the Indifferent cluster set as the reference category in both. Variables that reached a p-value < 0.30 in bivariate analyses were considered for inclusion, along with age and gender which were retained regardless of their bivariate significance. The final set of independent variables for the asthma model comprised age, gender, income, additional tobacco use, stage of change, and nicotine dependence (FTND). The lung cancer model additionally included educational attainment and age at smoking initiation. Nicotine dependence, which is partly derived from daily cigarette consumption, showed a high correlation with the number of cigarettes smoked per day (ε² = 0.28, all pairwise rank-biserial r > 0.58, p < 0.001); therefore, to avoid multicollinearity, the latter variable was not entered into either model. For the lung cancer model, FTND was entered as a binary variable (Low vs. Moderate/High) to ensure model stability given the small number of participants with high dependence. Adjusted odds ratios (OR) and 95% confidence intervals (CI) were reported. All analyses were conducted using R version 4.4.0 and Jamovi version 2.4. A two-sided p-value of less than 0.05 was considered statistically significant.

Results

Participant characteristics

A total of 310 smokers from Khuzestan Province were enrolled, with the majority being male (n = 263, 84.8%) and nearly half holding a university degree (n = 143, 46.1%). About one-quarter reported no fixed income (n = 79, 25.5%), and nearly half used additional tobacco products, including hookah (n = 109, 35.2%) and electronic cigarettes (n = 36, 11.6%). Approximately 53.2% (n = 165) planned to quit within one month, while 35.2% (n = 109) reported no intention to quit within the next six months. Participants began smoking at a mean age of 19.2 years (SD = 6.9) and smoked an average of 12.2 cigarettes per day (SD = 8.7) (Table 1).

Table 1.

Participant demographics and tobacco use characteristics (n = 310)

		Count	Percent
Gender	Male	263	84.8
	Female	47	15.2
Educational attainment	No formal education / Primary	26	8.4
	Up to Diploma	141	45.5
	University/Academic degree	143	46.1
Income (million IRR/month)	No income	79	25.5
	Low income (< 8)	95	30.7
	Middle income (8–16)	85	27.4
	High income (> 16)	51	16.5
Additional tobacco use	Yes	147	47.4
	No	163	52.6
Stage of Change (Readiness to Quit Smoking)	Not planning to quit within 6 months	109	35.2
	Planning to quit within 6 months	36	11.6
	Planning to quit within 1 month	165	53.2
Nicotine Dependence Level	Low	188	60.7
	Moderate	81	26.1
	High	41	13.2
Age (years) Mean (SD)	28.8 (11.4)
Age at smoking initiation(years) Mean (SD)	19.2 (6.9)
Number of cigarettes smoked per day Mean (SD)	12.2 (8.7)

Values are presented as n (%) unless otherwise indicated. Continuous variables are reported as mean (SD). Stage of change reflects readiness to quit smoking based on the Transtheoretical Model

Perceived risk of asthma and lung cancer

The mean perceived susceptibility was 4.3 (SD = 3.3) for asthma and 4.7 (SD = 3.8) for lung cancer, while the mean perceived severity was 4.4 (SD = 3.5) for asthma and 5.3 (SD = 3.4) for lung cancer. Cluster analysis of perceived asthma risk revealed four distinct groups: the Indifferent cluster (Low/Low) included 25.2% of participants (n = 78); the Optimistic Bias cluster (Low/High) comprised 25.5% (n = 79); the Dismissive cluster (High/Low) included 24.8% (n = 77); and the Concerned cluster (Moderate/High) comprised 24.5% (n = 76) (Fig. 1A). Similarly, cluster analysis of lung cancer risk perception revealed four groups: Indifferent (Low/Low) included 39.7% (n = 123); Optimistic Bias (Low/High) included 16.5% (n = 51); Dismissive (High/Low) included 21.3% (n = 66); and Concerned (High/High) included 22.6% (n = 70), reflecting varied patterns of perceived risk (Fig. 1B).

Fig. 1.

Cluster distribution of perceived risk for asthma (A) and lung cancer (B) among adult cigarette smokers in Khuzestan Province, Iran (n = 310). Values represent mean scores on a 0–10 scale. Clusters were labeled as Indifferent, Optimistic Bias, Dismissive, and Concerned based on relative levels of perceived susceptibility and perceived severity. Sample sizes are shown for asthma and lung cancer, respectively

Comparison of participants’ characteristics across FTND nicotine dependence levels

Participants were categorized into low, moderate, and high nicotine dependence groups based on FTND classification (n = 232, 61, and 17, respectively). The characteristics of participants across FTND levels are presented in Table 2.

Table 2.

Comparison of participant characteristics across FTND-based nicotine dependence levels (n = 310)

Characteristic		Low N = 232a	Moderate N = 61a	High N = 17a	Overall p-valueb
Age		29 (11)	27 (9)	34 (20)	0.3
Cigarettes per day		10 (6)	17 (8)	30 (8)	<0.001
Smoking initiation age		20.0 (7.4)	16.9 (4.3)	15.8 (4.5)	<0.001
Gender	Male	191 (82)	55 (90)	17 (100)	0.058
	Female	41 (18)	6 (9.8)	0 (0)
Education	Low	15 (6.5)	8 (13)	3 (18)	0.043
	Diploma	101 (44)	30 (49)	10 (59)
	University	116 (50)	23 (38)	4 (24)
Income (million IRR/month)	No income	63 (27)	15 (25)	1 (5.9)	0.03
	Low income (< 8)	67 (29)	25 (41)	3 (18)
	Middle income (8–16)	67 (29)	10 (16)	8 (47)
	High income (> 16)	35 (15)	11 (18)	5 (29)
Other forms of tobacco use	No	126 (54)	29 (48)	8 (47)	0.6
	Yes	106 (46)	32 (52)	9 (53)

Post-hoc pairwise comparisons were conducted using Wilcoxon rank-sum test (continuous variables) and Chi-square tests with Bonferroni correction   

 p-values < 0.05 are shown in bold

^aValues are presented as mean (SD) or n (%)

^b Overall comparisons were performed using Kruskal–Wallis test for continuous variables and Pearson’s Chi-square test for categorical variables

Significant differences were observed in cigarette consumption and smoking initiation age across FTND groups. Post-hoc pairwise comparisons showed that cigarette consumption differed significantly between all groups (all adjusted p < 0.001). Smoking initiation age was significantly lower in the high dependence group compared to the low dependence group (p = 0.02), while no significant difference was observed between the high and moderate groups (p = 1.00); however, a significant difference was found between the moderate and low groups (p < 0.001).

No significant pairwise differences were observed in educational attainment after Bonferroni correction (High vs. Low: p = 0.053; High vs. Moderate: p = 0.55; Moderate vs. Low: p = 0.10). Similarly, for income level, a significant difference was found only between the high and moderate dependence groups (p = 0.014), while other pairwise comparisons were not significant (High vs. Low: p = 0.06; Moderate vs. Low: p = 0.14).

Associations between asthma risk perception and participant characteristics

In bivariate analyses (Table 3), asthma risk perception clusters differed significantly by gender, income, tobacco use, and stage of change. Specifically, the Optimistic Bias cluster had a lower proportion of women (6.3%) compared to the Indifferent cluster (21.8%) (p = 0.005), and participants in the Optimistic Bias cluster were more likely to report having no fixed income (40.5%) compared to other clusters (p = 0.03). Stage of change differed significantly across clusters (p = 0.002), with lower readiness to quit being more prevalent in the Indifferent and Dismissive clusters. Nicotine dependence level also differed significantly (p = 0.004); the Concerned cluster showed the highest proportion of high nicotine dependence (11.8%), whereas the Indifferent cluster had the lowest (2.6%).

Table 3.

Associations between asthma risk perception clusters and sociodemographic/tobacco use variables

		Asthma Clusters				pvalue	Post hoc pairwise comparisons (significant P-values)
		Indifferent (C1)	Optimistic Bias (C2)	Dismissive (C3)	Concerned (C4)
	Susceptibility	Low	Low	High	Moderate
	Severity	Low	High	Low	High
Gender	Male	61(78.2)	74(93.7)	63(81.8)	65(85.5)	0.04	C1 vs. C2 = 0.005
	Female	17(21.8)	5(6.3)	14(18.2)	11(14.5)
Educational attainment	No formal education / Primary	9(11.5)	6(7.6)	4(5.2)	7(9.2)	0.58	-
	Up to Diploma	29(37.2)	36(45.6)	40(51.9)	36(47.4)
	University/Academic degree	40(51.3)	37(46.8)	33(42.9)	33(43.4)
Income (million IRR/month)	No fixed income	16(20.5)	32(40.5)	14(18.2)	17(22.4)	0.03	C1 vs. C2 = 0.001 C2 vs. C3 = 0.003 C2 vs. C4 = 0.004
	Low income (< 8)	23(29.5)	24(30.4)	26(33.8)	22(28.9)
	Middle income (8–16)	21(26.9)	16(20.3)	26(33.8)	22(28.9)
	High income (> 16)	18(23.1)	7(8.9)	11(14.3)	15(19.7)
Additional tobacco use	Yes	38(48.7)	25(31.6)	46(59.7)	38(50.0)	0.005	C2 vs. C3 < 0.001
	No	40(51.3)	54(68.3)	31(40.3)	38(50.0)
Stage of Change (Readiness to Quit Smoking)	Not planning to quit within 6 months	56(71.8)	37(46.8)	55(71.4)	40(52.6)	0.002	C1 vs. C2 = 0.004 C1 vs. C4 = 0.005 C2 vs. C3 = 0.005
	Planning to quit within 6 months	11(14.1)	26(32.9)	16(20.8)	28(36.8)
	Planning to quit within 1 month	11(14.1)	16(20.2)	6(7.8)	8(10.5)
Nicotine Dependence Level	Low	63 (80.77)	52 (65.82)	66 (85.71)	51 (67.11)	0.004	C2 vs. C3 = 0.048
	Moderate	13 (16.67)	24 (30.38)	8 (10.39)	16 (21.05)
	High	2 (2.56)	3 (3.80)	3 (3.90)	9 (11.84)
Age	Mean (SD)	30.4 (13.1)	26.1 (9.4)	28.6 (9.4)	30.2 (12.9)	0.1	-
Age at smoking initiation (years)	Mean (SD)	19.56(6.3)	19.1 (9.8)	19.4 (4.9)	18.6 (5.6)	0.35	-
Number of cigarettes smoked per day	Mean (SD)	10.7 (8.2)	13.7 (7.7)	10.1 (8.1)	14.3 (10.1)	< 0.001	C1 vs. C2 = 0.028 C2 vs. C3 = 0.004 C3 vs. C4 = 0.019

Values are presented as n (%), unless otherwise indicated

P-values derived from Chi-square tests for categorical variables and Kruskal–Wallis tests for continuous variables, with Bonferroni-adjusted post-hoc comparisons

In the multinomial logistic regression model with the Indifferent cluster as reference (Table 4), male gender independently predicted membership in the Optimistic Bias cluster (OR = 6.04, 95% CI: 1.86–19.57). Higher income (> 16 million IRR) significantly reduced the odds of Optimistic Bias membership relative to having no income (OR = 0.14, 95% CI: 0.04–0.52). Intention to quit within six months was associated with significantly increased odds of belonging to both the Optimistic Bias (OR = 2.90, 95% CI: 1.21–6.97) and Concerned (OR = 3.59, 95% CI: 1.56–8.26) clusters. Compared to smokers with low nicotine dependence, those with moderate FTND levels were more likely to be in the Optimistic Bias cluster (OR = 2.40, 95% CI: 1.02–5.61), whereas those with high FTND levels were more likely to be in the Concerned cluster (OR = 5.60, 95% CI: 1.11–28.34). Age and additional tobacco use were not significant predictors in the fully adjusted model.

Table 4.

Multinomial Logistic Regression Results for Asthma Risk Perception Clusters

Variable	Level	Optimistic Bias vs. Indifferent OR (95% CI)	Dismissive (C3) vs. Indifferent OR (95% CI)	Concerned (C4) vs. Indifferent OR (95% CI)
Age	(Continuous)	0.97 (0.94–1.01)	0.99 (0.95–1.02)	0.99 (0.96–1.03)
Gender	Female (Ref.)	1.00	1.00	1.00
	Male	6.04 (1.86–19.57)	1.48 (0.58–3.77)	1.65 (0.61–4.44)
Income (million IRR/month)	No fixed income (Ref.)	1.00	1.00	1.00
	Low income (< 8)	0.44 (0.17–1.12)	1.25 (0.48–3.30)	0.73 (0.28–1.94)
	Middle income (8–16)	0.35 (0.11–1.06)	1.39 (0.46–4.22)	0.78 (0.25–2.39)
	High income (> 16)	0.14 (0.04–0.52)	0.79 (0.23–2.75)	0.60 (0.18–2.00)
Additional tobacco use	Yes (Ref.)	1.00	1.00	1.00
	No	1.80 (0.87–3.74)	0.65 (0.33–1.28)	0.87 (0.43–1.73)
Stage of Change (Readiness to Quit Smoking)	Not planning to quit within 6 months (Ref.)	1.00	1.00	1.00
	Planning to quit within 6 months	2.90 (1.21–6.97)	1.55 (0.64–3.72)	3.59 (1.56–8.26)
	Planning to quit within 1 month	1.86 (0.71–4.86)	0.59 (0.20–1.76)	0.96 (0.34–2.69)
Nicotine Dependence (FTND)	Low (Ref.)	1.00	1.00	1.00
	Moderate	2.40 (1.02–5.61)	0.54 (0.20–1.43)	1.63 (0.69–3.83)
	High	2.43 (0.36–16.18)	1.43 (0.22–9.14)	5.60 (1.11–28.34)

Statistically significant odds ratios (95% CI does not include 1.00) are shown in bold

OR  Odds Ratio, CI  Confidence Interval, Ref. Reference category

Associations between lung cancer risk perception and participant characteristics

In bivariate analyses (Table 5), lung cancer risk perception clusters differed significantly by educational attainment (p = 0.003) and income (p < 0.001). The Indifferent cluster had a higher proportion of participants with university education compared to the other clusters. Regarding income, the Indifferent cluster had a notably higher proportion of participants with no fixed income (34.1%) compared to the other clusters, whereas the Dismissive and Concerned clusters were characterized by the highest proportions of low-income participants (< 8 million IRR) (43.9% and 47.1%, respectively). Stage of change also differed significantly across clusters (p = 0.002), primarily driven by the Optimistic Bias cluster (C2), which had a substantially lower proportion of smokers with no plan to quit within six months (33.3%) compared to the other three clusters (all > 62%). Nicotine dependence level varied significantly (p = 0.005), with the Optimistic Bias cluster showing the highest proportion of high nicotine dependence (11.8%). Daily cigarette consumption was significantly higher in the Concerned cluster (14.7 cigarettes/day) and lowest in the Indifferent cluster (9.4 cigarettes/day) (p < 0.001). Age also differed significantly across clusters (p = 0.005), with the Dismissive cluster being the oldest on average (33.2 years) and the Concerned cluster the youngest (27.1 years).

Table 5.

Associations between lung cancer risk perception clusters and sociodemographic/behavioral variables

		Lung cancer Clusters				pvalue	Post hoc pairwise comparisons (significant P-values)
		Indifferent (C1)	Optimistic Bias (C2)	Dismissive (C3)	Concerned (C4)
	Susceptibility	Low	Low	High	High
	Severity	Low	High	Low	High
Gender	Male	98(79.7)	46(90.2)	58(87.9)	61(87.1)	0.22	-
	Female	25(20.3)	5(9.8)	8(12.1)	9(12.9)
Educational attainment	No formal education / Primary	9(7.3)	4(7.8)	5(7.6)	8(11.4)	0.003	C1 vs. C2 = 0.001 C1 vs. C4 = 0.002
	Up to Diploma	39(31.7)	27(52.9)	37(56.1)	38(54.3)
	University/Academic degree	75(61.0)	20(39.2)	24(36.4)	24(34.3)
Income (million IRR/month)	No fixed income	42(34.1)	10(19.6)	12(18.2)	15(21.4)	< 0.001	C1 vs. C2 = 0.001 C1 vs. C4 < 0.001
	Low income (< 8)	22(17.9)	11(21.6)	29(43.9)	33(47.1)
	Middle income (8–16)	30(24.4)	23(45.1)	16(24.2)	16(22.9)
	High income (> 16)	29(23.6)	7(13.7)	9(13.6)	6(8.6)
Additional tobacco use	Yes	57(46.3)	21(41.2)	28(42.4)	41(58.6)	0.17	-
	No	66(53.7)	30(58.8)	38(57.6)	29(41.4)
Stage of Change (Readiness to Quit Smoking)	Not planning to quit within 6 months	83(67.5)	17(33.3)	44(66.7)	44(62.9)	0.002	C1 vs. C3 < 0.001 C2 vs. C3 = 0.002 C3 vs. C4 = 0.005
	Planning to quit within 6 months	25(20.3)	23(45.1)	14(21.2)	19(27.1)
	Planning to quit within 1 month	15(12.2)	11(21.6)	8(12.1)	7(10.0)
Nicotine Dependence Level	Low	106 (86.18)	35 (68.63)	45 (68.18)	46 (65.71)	0.005	C1 vs. C2(opt)=0.008 C1 vs. c3(Dis)=0.036 C1 vs. C4 = 0.006
	Moderate	16 (13.01)	10 (19.61)	17 (25.76)	18 (25.71)
	High	1 (0.81)	6 (11.76)	4 (6.06)	6 (8.57)
Age	Mean (SD)	28.1 (9.9)	28.6 (13.6)	33.2 (13.2)	27.1 (9.7)	0.005	C1 vs. C3 = 0.043 C2 vs. C3 = 0.004 C3 vs. C4 = 0.018
Age at smoking initiation (years)	Mean (SD)	19.4 (4.7)	18.2 (5.5)	19.2 (5.5)	19.6 (11.2)	0.1	-
Number of cigarettes smoked per day	Mean (SD)	9.4 (6.7)	13.6 (8.4)	13.7 (10.0)	14.7 (9.8)	< 0.001	C1 vs. C2 = 0.002 C1 vs. C4 < 0.001

Values are presented as n (%) or mean (SD). Lung cancer risk perception clusters were defined based on perceived susceptibility and perceived severity scores. P-values were obtained using Chi-square tests for categorical variables and Kruskal–Wallis tests for continuous variables, with Bonferroni-adjusted post-hoc comparisons

To identify independent predictors of cluster membership, a multinomial logistic regression model with the Indifferent cluster as reference was fitted (Table 6). Low income (< 8 million IRR) was also a significant independent predictor, increasing the odds of belonging to the Optimistic Bias (OR = 4.26, 95% CI: 1.70–10.68) and Concerned (OR = 3.69, 95% CI: 1.51–9.01) clusters. Intention to quit within six months or one month significantly increased the likelihood of Dismissive cluster membership (OR = 5.25, 95% CI: 2.22–12.43 and OR = 3.81, 95% CI: 1.35–10.76, respectively). Moderate or high nicotine dependence (FTND) independently predicted membership in all three clusters relative to the Indifferent profile: Optimistic Bias (OR = 2.69, 95% CI: 1.18–6.12), Dismissive (OR = 3.04, 95% CI: 1.24–7.45), and Concerned (OR = 3.21, 95% CI: 1.43–7.19). Age, age at smoking initiation, gender, educational attainment, and additional tobacco use were not significant predictors in the fully adjusted model.

Table 6.

Multinomial Logistic Regression Results for Lung Cancer Risk Perception Clusters

Variable	Level	Optimistic Bias vs. Indifferent OR (95% CI)	Dismissive (C3) vs. Indifferent OR (95% CI)	Concerned (C4) vs. Indifferent OR (95% CI)
Age	(Continuous)	1.02 (0.98–1.07)	1.04 (1.00–1.09)	1.00 (0.96–1.04)
Age at smoking initiation (years)	(Continuous)	0.97 (0.90–1.04)	0.97 (0.89–1.05)	1.02 (0.97–1.08)
Gender	Female (Ref.)	1.00	1.00	1.00
	Male	1.45 (0.53–3.92)	1.18 (0.35–4.02)	1.93 (0.73–5.16)
Educational attainment	No formal education / Primary (Ref.)	1.00	1.00	1.00
	Up to Diploma	2.08 (0.56–7.79)	1.98 (0.45–8.69)	1.23 (0.37–4.02)
	University/Academic degree	0.69 (0.18–2.62)	0.64 (0.14–2.86)	0.38 (0.12–1.26)
Income (million IRR/month)	No fixed income (Ref.)	1.00	1.00	1.00
	Low income (< 8)	4.26 (1.70–10.68)	2.15 (0.72–6.44)	3.69 (1.51–9.01)
	Middle income (8–16)	1.60 (0.55–4.71)	2.77 (0.89–8.65)	1.41 (0.49–4.03)
	High income (> 16)	0.56 (0.17–1.88)	0.47 (0.12–1.86)	0.39 (0.11–1.41)
Additional tobacco use	Yes (Ref.)	1.00	1.00	1.00
	No	1.29 (0.65–2.56)	0.96 (0.44–2.07)	0.65 (0.33–1.29)
Quit decision (Stage of change)	Not planning to quit within 6 months (Ref.)	1.00	1.00	1.00
	Planning to quit within 6 months	1.09 (0.48–2.49)	5.25 (2.22–12.43)	1.71 (0.77–3.79)
	Planning to quit within 1 month	1.12 (0.41–3.08)	3.81 (1.35–10.76)	1.16 (0.41–3.33)
Nicotine dependence (FTND)	Low (Ref.)a	1.00	1.00	1.00
	Moderate or High	2.69 (1.18–6.12)	3.04 (1.24–7.45)	3.21 (1.43–7.19)

Values are odds ratios (OR) and 95% confidence intervals (CI) from multinomial logistic regression with the Indifferent cluster as reference; Ref. = Reference category. Statistically significant ORs (95% CI excludes 1.00) are shown in bold

^aFTND was entered as a binary variable: Low (reference) vs. Moderate or High

Discussion

This study aimed to explore patterns of perceived risk for lung cancer and asthma among adult cigarette smokers in Khuzestan Province, Iran, using cluster analysis. By jointly examining perceived susceptibility and perceived severity, we identified four distinct groups of risk perception for each disease: Indifferent, Optimistic Bias, Dismissive, and Concerned. These clusters revealed differences in how smokers perceive their vulnerability to smoking-related health problems. They also varied across several sociodemographic characteristics, levels of nicotine dependence, and readiness to quit (stage of change).

In the asthma model, both the Concerned and Optimistic Bias clusters demonstrated greater readiness to quit compared to the Indifferent and Dismissive clusters. Notably, smokers in the Optimistic Bias cluster—despite perceiving low personal susceptibility—exhibited the most favorable readiness profile, with the lowest proportion in precontemplation (46.8%) and the highest proportion planning to quit within one month (20.2%). At the same time, the Concerned cluster displayed a notable paradox: its members carried higher nicotine dependence yet reported readiness to quit comparable to the Optimistic Bias group, with over 47% intending to quit within six months or one month. These patterns suggest that, at least for asthma, perceiving high disease severity, even in the absence of perceived personal susceptibility, can motivate concrete quit intentions—an insight with direct implications for message framing in cessation programs.

Unlike previous studies that relied on a single overall measure of perceived risk, this study examined perceived susceptibility and severity as distinct dimensions, allowing for a more nuanced description of smokers’ risk perceptions.

To our knowledge, this is among the first studies in Iran to apply cluster analysis to examine perceived risk of lung cancer and asthma among cigarette smokers, building on our earlier research on hookah smokers.

Several international studies have shown that smokers’ perceptions of risk are highly varied [40], and that those who perceive greater health risks often differ in their smoking behaviors and intentions to quit. For example, Chen et al. found that smokers who perceived themselves at greater risk of developing lung cancer tended to engage in fewer risky smoking behaviors and were more aware of smoking-related harms [41]. Similarly, Savoy et al. reported that higher perceived risk was associated with stronger intentions to quit smoking [42]. These findings are consistent with our results, in which participants in the Concerned cluster, who perceived high susceptibility and severity, showed greater readiness to quit despite higher nicotine dependence. This finding contrasts with some previous work that linked higher perceived risk with lower dependence, suggesting that the relationship between risk perception and nicotine dependence may be more complex than previously thought and likely depends on the specific profile of susceptibility and severity held by the smoker.

An additional and noteworthy observation in the present study relates to differences in cluster distribution between the two diseases. A notable finding in this study was the higher proportion of smokers classified in the Indifferent cluster for lung cancer risk perception (39.7%) compared with asthma risk perception (25.2%). This difference may reflect variations in how smokers cognitively evaluate different smoking-related diseases. Lung cancer is often perceived as a more distal and uncertain outcome, which may contribute to lower perceived personal relevance and increased psychological distancing, thereby increasing the likelihood of indifferent responses. In contrast, asthma is generally associated with more immediate respiratory symptoms and functional impairment, which may enhance personal salience and reduce the proportion of individuals with uniformly low perceived susceptibility and severity.

This pattern is consistent with previous evidence suggesting that smokers tend to underestimate long-term, life-threatening conditions such as lung cancer while showing relatively higher sensitivity to conditions with more immediate and experiential manifestations [40–42]. Moreover, optimistic bias and defensive processing of risk information have been shown to be particularly strong in relation to cancer-related outcomes compared to other chronic respiratory diseases [10, 18].

These findings suggest that tailoring risk communication strategies according to disease type may be beneficial. In particular, lung cancer-related messages may need to reduce psychological distance by emphasizing more immediate and tangible health consequences of smoking, whereas asthma-related communication may be more effective when focusing on long-term disease management and quality-of-life implications.

Similarly, Iranian studies have reported patterns that are broadly consistent with our findings, although they differ in their analytical approaches and interpretations. Recent research has shown that many current smokers exhibit only moderate levels of perceived risk for respiratory diseases, and that this perception alone does not necessarily translate into intentions to quit [10, 43]. Our results align with this pattern, particularly within the Indifferent and Dismissive groups, who remain at the precontemplation stage of quitting. Moreover, findings from Iranian studies on hookah users have emphasized the need for targeted educational and risk-awareness interventions, which closely correspond to the needs identified in our study population [43].

In most previous studies, analyses of perceived risk have typically relied on total or average scores (variable-centered). In contrast, our study employed cluster analysis to identify distinct risk-perception profiles among cigarette smokers, allowing for a more detailed examination of individual differences and subgroup characteristics. This approach has also been used by Wen et al. and Dijkstra et al. to identify smoker subgroups that differ in nicotine dependence, readiness to quit, and cognitive appraisals [29, 30]. For each condition, we identified four distinct clusters—labeled here as Indifferent, Optimistic Bias, Dismissive, and Concerned—which aligns with the findings of previous studies. By separating the dimensions of susceptibility and severity, these clusters revealed profiles such as Optimistic Bias (low-susceptibility/high-severity) and Dismissive (high-susceptibility/low-severity)—patterns that would not be detectable using single-score methods. These findings highlight that identifying specific smoker profiles is an important consideration in enhancing the impact of future interventions.

Beyond methodological considerations, in this study, the clusters varied in sociodemographic and behavioral characteristics, including income, education, additional tobacco use, nicotine dependence, and readiness to quit. Overall, these patterns were broadly consistent with those observed in previous research. Prior studies have shown that lower socioeconomic status and higher nicotine dependence are associated with lower perceived personal risk, whereas higher perceived risk is linked to greater motivation to quit—both among individuals receiving clinical care and participants in screening programs [29, 30, 41, 44].

However, the multinomial logistic regression analyses, which adjusted for all covariates simultaneously, provided a more refined and somewhat counterintuitive picture. For lung cancer, smokers with low nicotine dependence were more likely to belong to the Indifferent cluster, whereas those with moderate or high dependence combined had significantly higher odds of membership in the other three clusters. Additionally, low income (< 8 million IRR) independently increased the likelihood of being in the Optimistic Bias or Concerned clusters. Notably, age and educational attainment—which were significant in bivariate tests—were no longer significant after adjustment, suggesting that their effects are mediated by dependence or income. For asthma, male gender remained a robust independent predictor of the Optimistic Bias profile, and intention to quit within six months independently predicted both the Optimistic Bias and Concerned profiles, whereas income and additional tobacco use lost their bivariate significance in the adjusted model. Importantly, the finding that the Indifferent cluster was independently characterized by low rather than high nicotine dependence suggests that a lack of perceived risk in this group is not simply a by-product of severe addiction. Instead, this cognitive profile may stem from other unmeasured factors, such as low health literacy or limited exposure to anti-smoking messages. Conversely, the Concerned cluster was associated with higher nicotine dependence and, compared to the Indifferent and Dismissive groups, greater readiness to quit, highlighting a subgroup of smokers who, despite physical addiction, acknowledge their vulnerability and express motivation to change. These nuanced patterns indicate that coupling generic information about the health risks of tobacco use with practical cessation support and tailored dependence-reduction strategies may improve the effectiveness of public health messages in Khuzestan, particularly for those who remain cognitively disengaged from the risks of smoking.

Study strengths

This study has several notable strengths. As one of the first studies in Iran to apply a person-centered approach to examining perceived risk of both lung cancer and asthma, it provides novel insights into smokers’ risk perceptions. By analyzing perceived susceptibility and severity separately, the study offers a more nuanced understanding than traditional variable-centered approaches. The identification of distinct subgroups enables targeted insights for public health interventions. A wide range of sociodemographic and behavioral factors—including age, gender, education, income, nicotine dependence, and readiness to quit—was considered, strengthening the understanding of factors influencing risk perception. Recruiting 310 adult cigarette smokers from multiple cities enhanced the generalizability of findings within the region. Finally, validated instruments were used for assessing risk perception, nicotine dependence (Fagerström Test), and stage of change, improving the reliability and validity of results.

Study limitations

Several limitations should be noted. The cross-sectional design prevents causal inferences between perceived risk and smoking behaviors. Convenience sampling may introduce selection bias, limiting generalizability. Self-reported data may be influenced by recall or social desirability bias, particularly regarding smoking behaviors and intentions to quit. Biochemical verification of smoking status was not included, which could have strengthened behavioral validity. The study focused primarily on cigarette smokers; while dual users were included, exclusive users of other tobacco products, such as hookah or e-cigarettes, were excluded. Finally, potentially influential factors, including mental health status, cultural beliefs, and exposure to public health campaigns, were not assessed. Despite these limitations, the study provides valuable insights into risk perception profiles, which can guide targeted smoking cessation strategies in Iran. Finally, due to the relatively modest overall sample size, several adjusted odds ratios—although statistically significant—were accompanied by wide confidence intervals, reflecting limited precision in certain subgroup comparisons. These estimates should be interpreted with caution and warrant replication in larger, independent samples.

Conclusion

This study identified four distinct risk-perception subgroups—Indifferent, Optimistic Bias, Dismissive, and Concerned—among cigarette smokers in Khuzestan Province based on perceived susceptibility and severity of lung cancer and asthma. Multinomial logistic regression revealed that the Indifferent cluster was independently characterized by low nicotine dependence, whereas the Concerned cluster combined higher dependence with greater readiness to quit. The Optimistic Bias cluster, despite low perceived personal susceptibility, displayed the most favorable quit-readiness profile in the asthma model, suggesting that perceived severity, even in the absence of personal susceptibility, may serve as a strong driver of quit intentions. Low income and moderate or high nicotine dependence emerged as robust predictors of cluster membership for lung cancer, while male gender and high income independently predicted specific asthma clusters. These heterogeneous profiles underscore the need to move beyond uniform risk communication. Tailored interventions should address underlying barriers to risk acknowledgment in Indifferent smokers, leverage high severity perception in Optimistic Bias smokers, and combine cessation support with dependence treatment for Concerned smokers.

Future directions

Future research should consider longitudinal designs to examine causal links between risk perception and smoking behaviors over time. Expanding research to include users of other tobacco products, such as hookah and e-cigarettes, would provide a more comprehensive understanding. Studies in diverse cultural and regional contexts within Iran and the broader Middle East would help assess generalizability. Incorporating factors such as mental health, exposure to anti-smoking campaigns, and cultural beliefs could further clarify determinants of risk perception and support the development of more targeted cessation interventions. Mixed-methods approaches could provide qualitative insights into underlying beliefs that quantitative surveys may not capture.

Authors' contributions

Zahra Shojaei: Conceptualization and design of the study, as well as the acquisition and collection of data. Zahra Khasraji: Conceptualization and design of the study, as well as the acquisition and collection of data. Abdurrhman Charkazi: Conceptualization and design of the study, Interpretation of data and substantively revised the manuscript. Bagher Pahlavanzadeh: Conceptualization and design of the study, as well as the analysis and interpretation of data; have drafted the work and revised.

Funding

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Data availability

Data is available from the corresponding authors on reasonable request.

Declarations

Ethics approval and consent to participate

The study protocol was approved by the Ethics Committee of Abadan University of Medical Sciences (Approval number: IR.ABADANUMS.REC.1402.064) and was conducted in accordance with the principles of the Declaration of Helsinki. Informed consent was obtained from all participants prior to their inclusion in the study. Participants were assured of anonymity and confidentiality and were provided with a comprehensive explanation of the study objectives before completing the questionnaire.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.