Psychological distress in urban Iran: a multilevel analysis of socioeconomic position and context

Abstract

Background

Mental health disorders significantly contribute to the global burden of disease, both as non-communicable diseases (NCDs) and as risk factors for other health conditions. They represent a major public health concern in low- and middle-income countries (LMICs). This study aims to investigate associations between psychological distress, socioeconomic factors, and corruption-related factors among Iranian adults in urban settings using multilevel analysis.

Methods

This cross-sectional analysis uses data from a survey conducted in 2025 among urban adults across all 31 provinces of Iran, employing a proportional, stratified cluster sampling design with probability proportional to size (PPS). Psychological distress among adults aged 18 years and above was assessed using the 12-item General Health Questionnaire (GHQ-12), with a binary threshold of ≥ 4. Multilevel logistic regression models were used to examine associations between psychological distress and socioeconomic factors, perceived corruption, and experience of paying bribes. Models adjusted for health status and included random effects at the province and county levels. Sensitivity analyses were conducted using alternative GHQ-12 thresholds (≥ 3 and ≥ 5) to assess the robustness of findings to the choice of cut-off.

Results

The proportion of participants scoring above the GHQ-12 threshold of ≥ 4 for psychological distress was 41.2% (mean = 3.46, SD = 3.54). Psychological distress showed clustering at both province and county levels. In the fully adjusted model, compared with married participants, single (OR = 1.32, 95% CI: 1.21–1.44) and divorced/separated participants (OR = 1.48, 95% CI: 1.26–1.74) reported higher odds of psychological distress, whereas widowed participants (OR = 1.18, 95% CI: 0.96–1.43) showed no statistically significant difference. Higher education (OR = 0.80, 95% CI: 0.72–0.89) and better subjective socioeconomic status (OR = 0.30, 95% CI: 0.25–0.37 for highest vs. lowest category) were associated with lower odds. Compared with unemployment, all employment categories were associated with lower odds of psychological distress (ORs = 0.65–0.70, all p < 0.001). Poorer health status showed a graded association with higher odds of distress (ORs = 1.30–1.70). After adjustment for socioeconomic and health factors, perceived corruption was weakly but positively associated with psychological distress (OR = 1.024, 95% CI: 1.019–1.028), and experience of paying bribes was associated with higher odds of psychological distress (OR = 1.38, 95% CI: 1.27–1.51).

Conclusions

The findings indicate that participants’ psychological distress is associated with socioeconomic factors, perceived corruption, and experience of paying bribes, with variation across provinces and counties in urban Iran. These results underscore the relevance of socioeconomic disadvantage, corruption-related conditions in public services, and area-level contextual variation as important factors in understanding psychological well-being.

Introduction

Mental disorders substantially contribute to the global burden of disease, functioning both as non-communicable diseases (NCDs) and as risk factors for other health conditions [1–3]. Mental health is a crucial component of overall health, alongside physical and social dimensions. The World Health Organization (WHO) defines mental health as “a state of well-being in which the individual realizes his or her own abilities, can cope with the normal stresses of life, can work productively and fruitfully, and is able to make a contribution to his or her community” [4]. The United Nations Sustainable Development Goals (SDG 3.4) include the promotion of mental health and well-being as a key target for improving global health [5, 6].

While high-income countries report high rates of mental health disorders, low- and middle-income countries (LMICs) face additional challenges, including limited mental health infrastructure, inadequate services, and social stigma [2, 7]. Evidence indicates that people’s mental health and well-being are influenced by a wide range of biological, social, and environmental factors. Socioeconomic disadvantage, social isolation, and low social cohesion have consistently been associated with poorer mental health outcomes although the magnitude of these associations may vary across settings [2, 4, 8, 9]. According to the social determinants of health (SDH) framework, an individual’s socio-economic position creates differences in exposure and vulnerability to health-compromising conditions. This social standing is shaped by the broader socio-economic and political context, which generates social stratification and hierarchy through policies, governance structures, and cultural norms [9]. Research from multiple disciplines suggests that mental health is influenced by the interaction of individual experiences and skills with social and economic circumstances. Understanding these relationships is important for developing effective interventions that address both individual- and structural-level determinants of mental health [4, 6, 9].

Iran has undergone epidemiological transitions over recent decades, during which the burden of mental disorders has increased [10–12]. Recent Global Burden of Disease (GBD)-based studies indicate that Iran’s burden of mental disorders has risen over recent decades with much of the increase reflecting demographic change rather than a large rise in age-adjusted burden. The burden also varies across provinces, pointing to subnational differences in risk and need [12]. National surveys further highlight these disparities: the 2021–2022 National Mental Health Survey reported a 12-month prevalence of 25.1% for psychiatric disorders, with marked differences by marital status, education, employment, and socioeconomic position [13]. Earlier surveys also reported an upward trend in psychological distress, particularly among urban residents and socially vulnerable groups [14]. Together, these findings point to an uneven mental health burden in Iran, with higher risk concentrated in urban residents and socially disadvantaged groups. The surveys also support a cautious interpretation of national trends as reflecting both epidemiological and social patterning rather than a uniform distribution.

As discussed, according to SDH approach, mental health is shaped by broader sociopolitical, economic, and cultural contexts rather than by individual characteristics alone [4, 6, 9]. Consistent with this perspective, the present study first considers individual socioeconomic characteristics in relation to psychological distress. Building on this, we further examine corruption-related factors that may be associated with psychological distress. These include perceived corruption and direct experiences of corruption through paying bribes for public services. Together, these measures provide a broader assessment of both contextual and socioeconomic factors.

Therefore, this study aimed to examine the socioeconomic gradient in psychological distress and whether perceived corruption and bribe payment for public services were associated with psychological distress after accounting for individual-level demographic, socioeconomic, and health-related factors, as well as contextual variation across provinces and counties.

Methods

This study analyzed data from a cross-sectional social health survey conducted in 2025 among 21,955 adults aged 18 years and older residing in urban areas across all 31 provinces of Iran.

Measures

The outcome variable, psychological distress, was assessed using the 12-item General Health Questionnaire (GHQ‑12) [15, 16]. Items were scored using the binary (0-0-1-1) method, resulting in a total score ranging from 0 to 12. A score of ≥ 4 was used to define a binary outcome of probable psychological distress [15, 17].

Independent variables comprised demographic factors (age, sex, and marital status), education, employment status, subjective household socioeconomic status, health status, perception of corruption, and experience of paying bribes for public services.

The subjective household socioeconomic status (SES) was assessed using the MacArthur Scale of Subjective Social Status [18]. Health status was assessed using a single item asking participants to evaluate their current overall health status (“All in all, how would you describe your current health status?”). Response options included: (1) healthy; (2) recent illness (≤ 6 months); (3) long-term illness (> 6 months); and (4) chronic condition or disability (physical or mental).

Perceived corruption was measured using a six-item scale. Respondents rated the extent of corruption in various societal institutions on a 6-point scale ranging from “None at all” to “Very high.” Item scores were summed to yield a total score ranging from 0 to 30, with higher scores indicating greater perceived corruption. The perceived corruption measure showed good internal consistency (Cronbach’s α = 0.91) and adequate sampling adequacy (KMO = 0.895), with exploratory factor analysis indicating a one-factor solution explaining 62.7% of the variance.

Experience of paying bribes for public services in the past year was assessed using a single-item question with response options of “Yes,” “No,” and “Prefer not to answer.” Responses were dichotomized, and participants who selected “Prefer not to answer” were excluded from the analysis.

Continuous variables (age and perceived corruption) were grand-mean centered before inclusion in the analyses.

Sampling methods

Participants were selected using a proportional, stratified cluster sampling design with probability proportional to size (PPS) to ensure representativeness of the urban population. Sampling strata were defined by province, county, and demographic composition (age and sex).

The sample was proportionally allocated to the 31 provinces based on their population aged 18 and above. Within each province, 3–5 counties were selected using PPS, resulting in 107 participating counties. Within selected counties, participants were selected from urban neighborhoods, proportionally distributed by sex and age group. In the final stage, within selected neighborhoods, a quota sampling strategy based on street intercept was employed. Trained interviewers approached individuals in public spaces (e.g., streets, squares, and parks) and residential areas. Participants were screened for eligibility (residency in the specific neighborhood and age requirement) and recruited to fulfill the pre-determined age and sex quotas for each county. This method was chosen to ensure the demographic composition of the sample matched the census data, while optimizing logistical feasibility [19, 20]. Data collection occurred in two daily time windows (10:00–13:00 and 16:00–20:00) to capture variation in participant availability.

Data collection

Data were collected via paper-based, self-administered questionnaires. Trained data collectors were present to provide assistance and retrieve the completed questionnaires, ensuring methodological consistency. Verbal informed consent was obtained from all participants prior to data collection, following a full explanation of the study’s purpose, confidentiality of their responses, and the right to withdraw.

Statistical analysis

Psychological distress was estimated using descriptive statistics. To examine factors associated with psychological distress, a binary outcome variable was defined based on the GHQ-12 score (≥ 4 indicating poor mental health versus < 4) [17, 19].

Prior to model estimation, multicollinearity among independent variables was assessed using variance inflation factors (VIFs). All VIF values were low (mean VIF = 1.82; maximum VIF = 3.39), indicating no problematic multicollinearity.

Given the hierarchical structure of the data, with individuals nested within counties and counties nested within provinces, multilevel logistic regression models with random intercepts were employed to account for clustering at both contextual levels. A preliminary null multilevel model with random intercepts at the province and county levels was first estimated to quantify between-area variation in psychological distress. The results indicated statistically significant variance at both the province and county levels, which supported the use of a multilevel framework and suggested that the majority of variation in psychological distress occurred at the individual level.

A range of available province-level indicators, including the Gini coefficient, GDP share, urban unemployment rates, average household income and expenditure, crime rates, and budget allocation, were initially examined as potential contextual covariates. However, preliminary analyses indicated that these variables did not meaningfully improve model fit; therefore, no higher-level covariates were included in the final models.

Subsequently, two sequential multilevel logistic regression models were estimated. Model 1 included individual socio-demographic characteristics (age, sex, education, marital status, and employment status), subjective socioeconomic status, and health status. Model 2 additionally incorporated corruption-related variables, including perceived corruption and experience of paying bribes, to examine their associations with psychological distress and to assess potential changes in the magnitude of the baseline covariates.

All models were estimated using mixed-effects logistic regression with random intercepts for province and county. Model estimates are reported as odds ratios with 95% confidence intervals.

As a robustness check, sensitivity analyses were conducted using alternative GHQ-12 cut-off points (≥ 3 and ≥ 5) to assess the stability of associations across different operational definitions of psychological distress. All analyses were conducted using Stata 15.1 (StataCorp, College Station, TX, USA).

Results

The participants’ mean age was 40.1 years (SD = 15.04), ranging from 18 to 84 years. Overall, 50.1% of respondents were female. The marital status distribution was as follows: married (57.8%), single (34.2%), divorced/separated (4.8%), and widowed (3.2%). Regarding subjective socioeconomic status, 31.6% of participants were in the two lowest quintiles (categories 1 and 2), 41.7% were in the middle quintile (category 3), and 26.7% were in the two highest quintiles (categories 4 and 5).

In total, 41.2% of participants scored above the GHQ-12 threshold for psychological distress (GHQ-12 ≥ 4). The mean GHQ-12 score for the sample was 3.46 (SD = 3.54).

The null multilevel logistic regression model demonstrated significant clustering of psychological distress across both provinces and counties. The likelihood ratio test provided strong evidence in favor of the multilevel model over a conventional single-level logistic regression (LR χ² = 596.48, p < 0.001). The intraclass correlation coefficients indicated significant clustering of psychological distress at both the province (ICC = 0.034) and county (ICC = 0.067) levels, suggesting meaningful variation across areas.

Model 1 indicated that, compared with married participants, being single (OR = 1.31, 95% CI 1.20–1.43, p < 0.001) and being divorced/separated (OR = 1.55, 95% CI 1.32–1.82, p < 0.001) were associated with higher odds of psychological distress, whereas widowhood was not significantly associated with the outcome. Higher education (university level) was associated with lower odds of psychological distress compared with the lowest educational level (OR = 0.84, 95% CI 0.75–0.93, p = 0.001). A socioeconomic gradient was observed, with higher socioeconomic categories associated with lower odds of psychological distress; individuals in the highest category had approximately 72% lower odds compared with those in the lowest category (OR = 0.28, 95% CI 0.23–0.35, p < 0.001). Compared with unemployment, all employment categories were associated with lower odds of psychological distress (ORs = 0.63–0.70, all p < 0.001). Poorer health status was associated with higher odds of psychological distress, ranging from OR = 1.34 for recent illness to OR = 1.78 for participants reporting disability (all p < 0.001) (Table 1).

Table 1.

Multilevel mixed-effects logistic regression models for psychological distress (GHQ-12 ≥ 4)

Variable	Model 1 OR (95% CI)	Model 2 OR (95% CI)
Age	0.992 (0.989–0.995) ***	0.993 (0.990–0.996) ***
Sex (ref: Male)
Female	1.07 (0.99–1.15)	1.11 (1.03–1.20) *
Marital status (ref: Married)
Single	1.31 (1.20–1.43) ***	1.32 (1.21–1.44) ***
Divorced/Separated	1.55 (1.32–1.82) ***	1.48 (1.26–1.74) ***
Widowed	1.19 (0.98–1.45)	1.18 (0.96–1.43)
Education (ref: Below high school diploma)
High school diploma	1.02 (0.91–1.14)	1.00 (0.90–1.12)
University education	0.84 (0.75–0.93) **	0.80 (0.72–0.89) ***
Socioeconomic status (ref: SES 1, Lowest)
SES 2	0.75 (0.65–0.85) ***	0.76 (0.66–0.86) ***
SES 3	0.49 (0.43–0.55) ***	0.51 (0.45–0.57) ***
SES 4	0.31 (0.27–0.35) ***	0.32 (0.28–0.37) ***
SES 5 (Highest)	0.28 (0.23–0.35) ***	0.30 (0.25–0.37) ***
Employment (ref: Unemployed)
Employed	0.70 (0.63–0.79) ***	0.70 (0.62–0.79) ***
Homemaker	0.65 (0.56–0.75) ***	0.66 (0.57–0.77) ***
Other	0.67 (0.58–0.78) ***	0.68 (0.59–0.80) ***
Retired	0.63 (0.53–0.76) ***	0.65 (0.54–0.78) ***
Health status (ref: Healthy)
Recent illness (≤ 6 months)	1.34 (1.18–1.53) ***	1.30 (1.14–1.48) ***
Long-term illness (> 6 months)	1.56 (1.39–1.75) ***	1.52 (1.35–1.70) ***
Chronic condition or disability	1.78 (1.48–2.15) ***	1.70 (1.40–2.05) ***
Perceived corruption	—	1.024 (1.019–1.028) ***
Bribe payment experience	—	1.38 (1.27–1.51) ***

p < 0.05*, p < 0.01**, p < 0.001***. OR  odds ratio, CI confidence interval

Model 2 showed that, after further adjustment for perceived corruption and bribery payment experience, the associations observed in Model 1 remained largely unchanged. Women showed slightly higher odds of psychological distress (OR = 1.11, 95% CI 1.03–1.20). Higher perceived corruption was associated with higher odds of psychological distress; however, the effect size was small, with each one-unit increase in the corruption score corresponding to a 2.4% increase in odds (OR = 1.024, 95% CI 1.019–1.028). Bribe payment experience was associated with 38% higher odds of psychological distress compared with no such experience (OR = 1.38, 95% CI 1.27–1.51), after adjustment for sociodemographic characteristics and health status (Fig. 1).

Fig. 1.

Adjusted odds ratios (ORs) and 95% confidence intervals from the final multilevel logistic regression model

Table 2 presents the random effects estimates and model fit statistics for the null model and multilevel models. Model fit improved with the sequential addition of predictors. Compared with the null model, Model 1 showed a substantial improvement in fit (LR χ² = 959.15, p < 0.001), accompanied by a reduction in AIC (21,015.11 to 20,091.96). The inclusion of corruption-related factors in Model 2 further improved model fit (LR χ² = 174.89, p < 0.001), with an additional decrease in AIC (19,921.07) and an improvement in log-likelihood (− 10,024.98 to − 9,937.54). Random intercept variances and ICCs changed only marginally across models but remained non-zero, indicating persistent geographic clustering in psychological distress after adjustment for individual-level covariates.

Table 2.

Random effects and model fit

Statistic	Null	Model 1	Model 2
Log likelihood	-10504.560	-10024.980	-9937.540
AIC	21015.110	20091.960	19921.070
BIC	21038.140	20253.180	20097.640
Province variance	0.118	0.105	0.113
County variance	0.119	0.121	0.121
ICC (Province)	0.034	0.030	0.032
ICC (County)	0.067	0.064	0.067
LR test vs. single-level model	χ²=596.48, p < 0.001	χ²=520.10, p < 0.001	χ²=528.61, p < 0.001

AIC Akaike Information Criterion, BIC Bayesian Information Criterion, ICC intraclass correlation coefficient, LR test  likelihood-ratio test comparing multilevel models with a single-level model. Lower values of AIC and BIC indicate better model fit

Sensitivity analyses

Sensitivity analyses were performed using alternative GHQ-12 cut-off points (≥ 3 and ≥ 5) to define psychological distress. The direction and statistical significance of the associations were broadly consistent across specifications, suggesting that the results were robust to the choice of threshold.

Discussion

This study analyzed data from a large-scale survey of the urban population across all 31 provinces of Iran. Overall, 41.2% of participants scored above the GHQ-12 cut-off (≥ 4), indicating probable psychological distress. The multilevel analyses demonstrated a consistent socioeconomic gradient, with lower socioeconomic position associated with higher odds of psychological distress. Health status was included as an important control variable and showed graded differences across categories. Marital status and other individual-level socioeconomic characteristics were also associated with psychological distress. Among the two corruption-related factors, and after adjustment for demographic, socioeconomic, and health status variables, having paid a bribe for public services was associated with higher odds of psychological distress, whereas perceived corruption showed a statistically significant but modest association, with a small effect size. Random-effects estimates indicated persistent between-province and between-county variation in psychological distress after adjustment for individual-level characteristics, suggesting residual contextual variation beyond the included covariates.

Our findings show an association between socioeconomic characteristics and psychological distress in this sample of the Iranian urban population, consistent with previous studies reporting similar associations between higher socioeconomic status and better mental health [21–25]. Longitudinal research indicates that both early-life socioeconomic disadvantage and social mobility trajectories influence mental health, though their relative importance remains debated [26, 27]. Prior multicohort analyses have reported that low socioeconomic status (SES) is linked to a higher risk of psychiatric disorders and subsequent chronic diseases [28], as well as higher mortality and a greater number of years of life lost, even after accounting for major behavioural and metabolic risk factors [29]. Consistent with the social determinants of health framework, disparities in mental health reflect systematic differences in socioeconomic position, and effectively addressing them requires integrated approaches that consider the complex interplay of biological, psychological, and social factors [8, 9, 30–32].

Our findings are consistent with prior evidence indicating that corruption, particularly direct experiences such as paying bribes for public services, is associated with poorer psychological well-being, although the strength of this association varies across contexts and study designs [33–36]. Service-specific evidence also suggests that bribery in health care and other public services may be linked to greater barriers to access and worse self-assessed health [33].Taken together, these findings support the interpretation of corruption as a potential psychosocial stressor, while the precise pathways remain to be established. It should be noted that bribery experience in this analysis was measured using a single dichotomous item, which does not capture the frequency, intensity, or contextual variation of corruption exposure; therefore, the results should be interpreted with caution.

While prior research from high-income countries has highlighted associations between structural conditions, socioeconomic disadvantage, and mental health [37–39], less is known about how these relationships operate in LMIC settings such as Iran. In contexts marked by economic instability and structural constraints, the pathways underlying psychological distress may differ from those observed in high-income countries, and future studies should examine whether distinct contextual mechanisms help explain these patterns. This is particularly important in global mental health, where unmet need, limited service capacity, and broader contextual barriers underscore the value of locally generated evidence for policy and intervention design [10, 40] .

This multilevel study may contribute to the literature in LMICs and suggests that efforts to address urban health disparities may benefit from considering both individual socioeconomic circumstances and broader contextual factors associated with well-being. Several limitations should also be acknowledged. First, our assessment was limited to the GHQ-12, a screening tool for psychological distress; therefore, the findings may not generalize to other mental health indicators or clinically diagnosed disorders. Second, the analysis was restricted to urban areas, which limits the generalizability of the results to the broader national population, including rural communities. Third, all study variables were based on self-reported data collected within a single survey, which may introduce common method bias. As noted in prior methodological literature, such designs can inflate observed associations due to shared measurement context rather than true underlying relationships [41]. In addition, health status was assessed using a self-reported measure, which may not fully capture objective or clinically verified health conditions. Given the sensitivity of topics such as bribery and corruption, responses may also be subject to social desirability bias and underreporting; however, anonymity procedures and self-administered questionnaires were used to help minimize these effects. Fourth, although the sampling design was proportionate at the provincial level and quotas were aligned with census age–sex distributions, the final street-intercept stage may have introduced selection bias by underrepresenting housebound individuals and overrepresenting those more present in public spaces. Data collection across two daily time windows was intended to reduce this bias, but it cannot eliminate it entirely. These limitations should be considered when interpreting the magnitude of the observed associations, and the findings should be interpreted as associative rather than causal.

Conclusions

The findings suggest a social gradient in psychological distress within this urban sample, with lower socioeconomic position associated with higher levels of distress across multiple indicators. Beyond individual-level characteristics, psychological distress was also associated with corruption-related factors, particularly the experience of paying bribes. These results underscore the relevance of socioeconomic disadvantage, corruption-related service experiences, and geographic variation in relation to psychological distress, suggesting the value of structurally informed and context-sensitive public mental health approaches in urban Iran. More broadly, these findings suggest a multi-level approach combining prevention, equitable service access, and social policy measures to address psychological distress and promote well-being.

Future work could explore how structural factors such as confidence in the public sector are linked to psychological distress. A key priority will be to investigate material circumstances, psychosocial stressors, social cohesion, and other contextual and individual-level conditions as interrelated dimensions rather than as isolated predictors. Longitudinal or mixed-method designs could help clarify temporal ordering and assess whether these associations differ by gender, socioeconomic position, or access to services. Comparative studies across urban and non-urban settings would further help identify which patterns are context-specific and which are more generalizable. Such work would better inform policy targets at both structural and service-delivery levels.

Abbreviations

GHQ-12

General Health Questionnaire (12-item)

LMICs

Low- and Middle-Income Countries

PPS

Probability Proportional to Size

SES

Socioeconomic Status

Authors’ contributions

F.A. and H.R.: Conceptualization. F.A., H.R., T.P., Y.S.: Methodology and data analysis. F.A., and SH.A.: Investigation.T.P. and F.A.: Drafted the manuscript. All authors reviewed and approved the final manuscript.

Funding

This study was supported by the Deputy for Research and Technology, Iran’s Ministry of Health and Medical Education (MOHME).

Data availability

The data are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

This study used data from a social health survey in Iran, which was approved by the Ethics Committee of the University of Social Welfare and Rehabilitation Sciences (Code: IR.USWR.REC.1402.068). The participants provided verbal consent. All methods/study procedures were carried out in accordance the 2024 Declaration of Helsinki. All participants were informed about the purpose of the study, assured of confidentiality, and provided verbal consent prior to participation. Participation was voluntary, and respondents could withdraw at any time without consequence.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.