Introduction

Obesity, as a critical public health concern, is one important risk factor and comorbidity of a wide range of health conditions, such as major depressive disorders, anxiety disorders, sleep disorders, cardiovascular diseases, kidney diseases, diabetes, neoplasms, cancers, all leading to an increased mortality risk compared to the overall population1,2,3,4,5,6,7,8. As estimated by the latest data, the global prevalence of obesity has increased from 8.8% (4.8%) in 1990 to 18.5% (14%) in 2022 in women and men9. With the increasing number of people with obesity, it is imperative to identity the potential risk factors to inform effective interventions. Food addiction (FA), as a promising risk factor, has attracted much attention during recent years both in the academia10,11,12,13,14,15,16,17 and by clinicians18.

As a phenotype which proposes that some people may develop addictive-like symptoms towards specific foods high in sugar and/or fat (especially ultra-processed foods19, such as chocolate and ice cream), FA may increase the food intake and contribute too overweight or obesity ultimately20,21. This is supported by the significant association found between FA and obesity in both children22, adolescents23 and adults24 in recent meta-analyses. Moreover, FA is closely related to the obesity-related constructs (such as impulsivity, reward sensitivity, loss of control, craving, cognitive dysfunction, mental distress, psychiatric comorbidities, social and occupational impairment)25,26,27,28,29,30,31,32, which further supports its vital role in the development of obesity. According to the latest meta-analysis, the pooled prevalence of FA is 15% among children and adolescents23, 14% for non-clinical adults33 and 40% in clinical samples, which is especially high in patients with eating disorders (bulimia nervosa (84%), binge eating disorder (63%) and anorexia nervosa (53%))34. Thus, the identification of associated factors of FA might hold great practical implications for the development of interventions to manage and decrease FA symptoms. Sex is a critical, though frequently neglected, candidate for the prevalence and the number of symptoms and the type of symptoms endorsed of FA.

Sex differences on FA

Sex is a vital dimension35, which calls for much more research attention in the field of addictive disorders36,37,38,39,40, and FA is no exception41. The sex differences of FA might be reflected on the prevalence, number of symptoms, specific symptoms endorsed and comorbidity and correlates.

As for the prevalence of FA, sex differences were repeatedly reported, with higher prevalence in female than male was found in meta-analysis of studies using the (Yale Food Addiction Scale (YFAS)42, while the latest meta-analysis of research employing YFAS 2.0 revealed that the prevalence of male (27%) is higher than that of female (24%)33. Meanwhile, no sex difference was also mentioned in several investigations43,44,45,46,47.

For the symptoms of FA, by using YFAS, no sex difference was revealed on the symptom score48 while sex differences were revealed for specific symptoms endorsed. Female students scored higher than male counterpart on “use despite knowledge of adverse consequences”, “persistent desire or repeated unsuccessful attempts to cut down”, “tolerance”, “withdrawal” and “clinically significant impairment” symptom whereas no sex difference was reported on “loss of control”, “much time spent to obtain, use, recover”, and “important activities given up” symptom49. Since four more symptoms were added into YFAS 2.0 when compared to YFAS, thus, it remains unknown whether there are any sex differences on these four symptoms.

For the comorbidity of FA, sex differences were frequently reported as well. The co-occurring FA and post-traumatic stress disorder (PTSD) was found to be higher for male (RR = 8.54) compared to female (RR = 4.32) after controlling for potential covariates in the community sample50, whereas this sex difference was absent in veterans51.

Relatively little attention was paid to the sex difference on the correlates of FA. The sex differences identified in this line of research demonstrated that sex-specific considerations were warranted when developing the interventions to alleviate the severity of FA. Chronotype (natural preferences of the body for wakefulness and sleep)52 was found to be a correlate of FA for males but not for females53. Specifically, being evening type might increase the risk of having a FA diagnosis only for men but not for women. Plasma dopamine levels was revealed to be negatively associated with food addiction in males, and positively in females54. For the adolescents seeking treatment for obesity, the sex differences about the attitudinal correlates of FA symptoms were revealed as well55. More specifically, functional impairment due to FA symptoms was associated with anxiety about gaining weight for the boys and social pressure about eating among the girls. Dieting was positively linked to avoidance from important social, functional and interpersonal activities due to eating among the boys, whereas it was correlated with continuing to eat despite knowing negative consequences among the girls. The association between social pressure about eating and eating for longer duration and higher amount than intended was only found for the girls but not for boys. Moreover, some sex-specific correlates of FA was also identified in previous research, such as polycystic ovarian syndrome for women56.

FA in Iran

The prevalence of FA in Iran ranged from 6.3% to 31% depending on the samples investigated47,57,58,59,60,61. The sex difference was only revealed in children and adolescent sample, with boys tending to be more likely to meet FA diagnosis than girls60. However, this sex difference was absent in Iranian adult samples47,57,58. Whether there are any potential sex differences on the comorbidity and correlates of FA, especially in Iran, remain largely unknown.

The current study

The previous studies conducted in Iran were limited by the small sample sizes (222 ~ 473 participants)47,57,58,59,60,61 and the specific research groups with health conditions (e.g., adults with obesity47,59), visitors of family physician clinics58 and individuals undergone laparoscopic sleeve gastrectomy57), which hinders from getting the prevalence of FA from the community level. In addition, the inclusion of single sex (such as only women were investigated in59) also unable to analyze the potential sex differences.

Thus, the current investigation aimed to determine the prevalence of in community level with large adult sample size in Iran and to explore whether there maybe potential sex-specific factors related to FA to help to develop more sex-appreciate interventions for FA in the future.

Methods

Design and participants

This cross-sectional study, conducted between 2021 and 2022, collected data from 9664 participants through an online questionnaire. The link of this online questionnaire was distributed by various social media, such as Twitter (X), Instagram, Facebook, and Telegram. Digital consent was obtained for each participant before formal questionnaires.

Measurements

The online questionnaire includes two parts. The first part is about participants’ basic demographic information which was designed mainly by referring previous literature62, such sex, age, and self-reported height and weight. Additionally, expanding previous studies, educational level, marital status, season of birth and the gambling participation las year (yes or no) were also collected to examine whether they were potential correlates of FA.

In adapting the scale to the Iranian cultural context, we also revised the examples of “Certain foods” that appear at the beginning of the YFAS 2.0. The original examples based on Western diets were replaced with culturally relevant food and beverage items commonly consumed in Iran. These included local equivalents of high-sugar, high-fat, and high-salt items to maintain the conceptual consistency of “highly palatable foods” while improving cultural sensitivity. Participants were instructed to reflect on any food or drink they personally struggled to control over the past year, in line with the scale’s conceptual framework.

In terms of psychometric properties, recent large-scale studies have confirmed the reliability and validity of both the Persian YFAS 2.0 and the 13-item modified version (mPYFAS 2.0). A study involving 9606 participants reported excellent model fit (CFI > 0.95, RMSEA < 0.03) and high internal consistency (α = 0.83, ordinal α = 0.93, CR = 0.86) 63. The full Persian YFAS 2.0 also showed strong construct validity and internal consistency in a separate non-clinical sample (KR-20 = 0.99, ω = 0.91) 64.

These procedures and findings support the use of the mPYFAS 2.0 as a valid and culturally appropriate measure for assessing food addiction in Iranian populations.

The second part is about the prevalence of FA, which was measured by Persian version of the modified Yale Food Addiction Scale 2.0 (mPYFAS 2.0). In the mPYFAS 2.0, 13 items are included to measure the status of FA according to the extrapolation of the DSM-5 criteria for substance use disorder to the consumption of highly palatable foods. The participants were asked to answer on a 8-point Likert scale ranging from “0” (Never) to “7” (Daily). The mPYFAS 2.0 showed satisfactory reliability and the sex invariance was also confirmed.

Statistical analysis

Data were carefully checked and cleaned before formal analysis. Data with missing responses, incomplete responses and careless responses were treated as invaid data and discarded. Descriptive results, such as frequency, percentage and mean, were calculated. Multicollinearity test was conducted to see if the current dataset was suitable for regression analysis. The results revealed that the VIF value of all variables was less than 2 (see the supplementary material), indicating that no multicollinearity was found among variables and the current dataset fulfilled the assumptions of regression analysis. Univariable logistic regression model was used to determine the associated variables of food addiction, which were employed to construct a multivariable logistic regression model to determine the correlates of food addiction after adjusting associated variables. The SPSS version 22 software (IBM Corporation, Armonk, NY, USA) was used for all the analysis.

The current investigation was conducted in accordance with the Declaration of Helsinki and the Ethical Guidelines for Medical and Health research established by ministry of Health and Medical Education and Ministry of Science, Research and Technology, Iran. The approval was obtained from the Ethics Review Committee of Shahroud Medical University Iran (Registration No. IR.SHMU.REC.1399.161). Some parts of the data were used for the psychometrics test of mPYFAS 2.0, which was published elsewhere.

Results

Participant characteristics

The data from 9664 participants was collected and data from 319 participants (3.3%) was discarded due to either incomplete responses or careless responses, which leading to 9345 valid datasets. The mean age of the 9345 participants included in the formal analysis was 29.67 ± 9.02 years, ranging from 18 to 73 years. The majority of the participants were female (82.10%). Nearly half participants (48.5%) belong to the normal weight category. And over half participants were single (53.9%) and got university degree (64.0%). Most respondents (96.1%) did not made any kind of online and offline bets or any kind of gambling during the last year (see Table 1).

Table 1 Sociodemographic characteristics of participants.
Full size table

Prevalence of food addiction

The diagnostic threshold for food addiction was met by 15.6% (95% CI [14.89%, 16.36%]) of participants (n = 1459), with 8.7% (95% CI [7.43%, 10.09%]) (n = 149) for male sample and 17.2% (95% CI [16.35%, 18.04%]) (n = 1310) for female sample. As for the food addiction severity, 4.0% (95% CI [3.62%, 4.42%]) met mild FA (male: 2.9% (95% CI [2.16%, 3.75%]); female: 4.3% (95% CI [3.83%, 4.74%])), 4.1% (95% CI [3.74%, 4.54%]) met moderate FA (male: 2.0% (95% CI [1.74%, 2.82%]); female: 4.6% (95% CI [4.14%, 5.08%])), and 7.5% (95% CI [6.97%, 8.04%]) met severe FA (male: 3.8% (95% CI [2.98%, 4.79%]); female: 8.3% (95% CI [7.73%, 8.97%])).

Table 2 Endorsement of individual food addiction symptoms.
Full size table

Symptoms of food addiction

The symptom score of female participants (1.58 ± 2.37) was significantly higher than that of male participants (1.16 ± 1.88), t = 7.79, Cohen’s d = 0.20. No sex difference was revealed on “important activities given up” symptom, and male participants endorsed more on “use in physically hazardous situations” and “continued use despite social or interpersonal problems” symptoms. Female participants endorsed more on the remaining symptoms and reported higher percentage to meet “clinical significant impairment” than male participants (see Table 2).

Correlates of food addiction

For the whole sample, the results of univariable logistic regression model showed that age, sex, weight status, education and marital status were associated with the FA diagnosis. However, only the associations with age, sex, BMI and education survived multivariable adjustment, while the association with marital status was rendered non-significant (see Table 3). Age (older than 30), being male, underweight/normal weight and high education level were negatively associated with FA.

Table 3 Logistic regression results for univariable analysis and multivariable analysis for FA diagnosis.
Full size table

To further explore the potential sex difference on correlates of FA, logistic regression was conducted for both females and males separately. For female sample, only age, weight status and marital status were identified as significant correlates of FA in univariable logistic regression model. After adjustment, age and weight status remained significant while marital status became insignificant (see Table 4). Age (22–25 years) and overweight and obesity were positively associated with FA for female participants.

Table 4 Logistic regression results for univariable analysis and multivariable analysis for FA diagnosis in female sample.
Full size table

For male sample, weight status, education and marital status were recognized as significant associates with FA in univariable logistic regression model while only weight status and education survived after adjustment in multivariable logistic regression model (see Table 5). Normal weight and high education level were negatively associated with FA for male participants.

Table 5 Logistic regression results for univariable analysis and multivariable analysis for FA diagnosis in male sample.
Full size table

Discussion

The present study represents the inaugural survey assessing the prevalence of food addiction and its associated demographic factors within a large non-clinical adult population of Iran. The sex differences were revealed on the prevalence, symptoms endorsed, and correlates of FA.

The prevalence of FA identified in the current study is 15.6%, which is exactly identical as Schulte and her colleagues’ finding during COVID-19 in USA64 and is also similar to the finding in Poland during COVID-19 (14.1%)65 and the latest meta-analysis data in non-clinical samples (14% around the world33 and 15% in in Latin America66). This prevalence is slightly lower than the prevalence (21%) of FA determined by a meta-analysis during COVID-1967, and much lower than the findings in Italy (46.9%)68 and Turkey (40.6%)69 while much higher than the prevalence in Bangladesh (7.5%)70. These inconsistencies might be due to several reasons. The YFAS version might be one of them. For example, Sultana et al.,70 used mYFAS, which might underestimate the prevalence of FA71. The sample characteristics might be another contributor. The current study recruited adult sample as Panno et al.,68 did, while Sultana et al.,70 and Kayaoğlu et al.,69 recruited only university students. However, even both employed university students as target sample69,70, noticeable difference of prevalence still was reported (40.6% vs. 7.5%). This is the same case for the current findings and Panno et al.68, which suggest that the prevalence of FA is heterogeneous across studies and some other variables might be more important than sample characteristics to lead to these discrepancies among previous studies, which warrant further investigations.

As for the sex difference on the prevalence of FA, we found that the prevalence of FA in Iran was more prevalent in women than men, which replicated previous findings42,62,72. More specifically, females got almost two-fold chance to get diagnosed as FA than males did (adjusted OR 2.94 (95%CI [2.43, 3.54]). The increased prevalence of FA in women as opposed to men may be associated with sex-related differences in hormonal profiles and/or dietary habits73,74.

Consistent with most previous findings24, individuals with overweight and obesity were susceptible to FA for both sex. Interestingly, the results for underweight represented opposite direction for females and males. Specifically, underweight was negatively associated with FA for females while positively associated with FA for males. These results suggested that the association between weight status and FA might be more complexed as previously assumed. In fact, Wiss75 summarized the previous studies about weight status and FA and found the heightened prevalence among the underweight population, which revealed a J-shaped relationship between BMI and FA prevalence43,76,77. We calculated the prevalence of FA for each status and found this J-shaped relationship was only revealed for male participants (see Table 6) but not female participants, which provides a more nuance insight about the linkage between weight status and FA prevalence and calls for special attention to underweight males.

Table 6 The prevalence of FA for different weight status.
Full size table

Notably, sex-specific association with FA was identified in the current study, which extends previous studies53,54,55,56. More specifically, age range (22–25 years) was found to be positively associated with FA for females, while education level (≥ high school) was negatively associated with FA for males was reported. The interventions targeting FA are still on its infancy and much endeavor is warranted78,79,80, however, the sex-specific associations identified in both the current study and previous ones cannot be neglected when developing interventions in the future.

The findings of the current study indicate that individuals in the younger age group (< 30 years) are more susceptible to food addiction. Previous studies have also demonstrated that younger individuals tend to score higher on symptoms of FA compared to their older counterparts43,81. This result may be interpreted as a component of heightened vulnerability to addictive behaviors in younger individuals, possibly due to limitations in their risk assessment abilities and potential difficulties in controlling resistance to the consumption of high-fat/high-sugar foods. Additionally, lifestyle and dietary habits undergo more significant changes in younger individuals, potentially exposing them to a greater risk of emotionally driven eating and susceptibility to stress-related factors82.

Consistent with previous literature49, no sex difference was revealed on “important activities given up” symptom, females endorsed more on “use despite knowledge of adverse consequences”, “persistent desire or repeated unsuccessful attempts to cut down”, “tolerance”, “withdrawal” and “clinically significant impairment” symptoms than males. However, we found females also endorsed more on “loss of control” and “much time spent to obtain, use, recover” symptoms, which did not agree with Yu et al.49. Expanding Yu et al.49, we found males endorsed more on “use in physically hazardous situations” and “continued use despite social or interpersonal problems” symptoms than females and female endorsed more on “craving” and “impaired daily functioning (work, school, home)” symptoms than males.

This observed sex-specific symptom pattern where females more frequently reported withdrawal, craving, and clinically significant impairment may reflect underlying psychological and sociocultural mechanisms. Although these symptoms are part of the mYFAS 2.0 diagnostic structure based on DSM-5 criteria, their expression may be influenced by emotional vulnerability and gendered stress exposure. Prior studies suggest that women are more likely to experience internalizing symptoms such as anxiety and depression, which are linked to emotionally driven eating behaviors83. Sociocultural pressures related to caregiving, appearance, and dual work-family roles may further exacerbate this pattern, making females more prone to affectively charged food addiction symptoms84,85. In contrast, males may exhibit more externally oriented symptoms such as tolerance and increased time spent, reflecting reward-driven patterns83. These findings underscore the need for gender-sensitive interventions that address both the neurobehavioral and contextual stress factors contributing to food addiction.

The cultural context of food in Iran significantly shapes how food addiction symptoms are experienced, especially among women. Food is deeply embedded in social and familial roles often managed by women heightening its emotional significance and the distress linked to perceived loss of control85,86,87. At the same time, ideals of thinness and restraint, particularly in urban areas, often conflict with food-centric traditions, fostering guilt and shame around eating88. This tension is intensified by Iran’s nutrition transition, where ultra-processed foods increasingly coexist with traditional high-calorie dishes, creating mixed health messages86,89. Religious and moral values around moderation can further compound emotional eating and shame87. These culturally specific dynamics suggest that DSM-5-based food addiction symptoms may manifest differently across settings and call for more context-sensitive assessment and intervention strategies85.

Given the complex nature of food addiction, effective interventions must address both psychological factors and sociocultural influences. Evidence-based approaches such as impulse-control training, cognitive-behavioral therapy (CBT), and mindfulness techniques have shown promise in treating compulsive eating90,91. Our findings, which revealed higher symptom severity among women, highlight the need for gender-sensitive interventions focused on emotional regulation and shame resilience87,88,92. On a broader scale, public health strategies should address the food environment by regulating ultra-processed food marketing, promoting food literacy, and encouraging healthy traditional diets89. Community-based programs that reflect cultural values around caregiving, body image, and moral expectations are essential for reducing stigma and enhancing relevance85,87. Integrating clinical, cultural, and policy-level strategies offers a comprehensive framework for prevention and treatment aligned with the drivers identified in this study.

Limitations and future directions

This study has several limitations. First, the sample included a disproportionately high number of female participants (82.1%), which may limit the generalizability of the findings particularly the sex-stratified interpretations. Although this reflects broader trends in voluntary survey participation and food addiction studies, it remains a noteworthy limitation. Second, the use of self-report measures introduces the possibility of social desirability and recall biases, particularly for sensitive behavioral symptoms. Third, the cross-sectional design precludes causal inferences regarding the relationship between sex-specific variables and food addiction symptoms. Fourth, the 15.6% prevalence of food addiction observed in our study is comparable to the weighted average of 20% reported in a recent meta-analysis of 272 studies using the YFAS and mYFAS33, this suggests our estimate is within the expected global range for general population samples. However, given that the diagnosis was based on self-administered questionnaires, some overestimation cannot be ruled out. Future studies using structured clinical interviews are needed to validate these rates. Additionally, while recent events such as the COVID-19 pandemic may have influenced eating behaviors, our study did not collect relevant temporal data to examine this impact directly. Future longitudinal research could explore how such contextual factors affect food addiction symptoms over time. Lastly, because the study was conducted in an Iranian cultural setting, with its own food practices, gender roles, and religious or moral expectations surrounding eating, the findings may not be directly generalizable to other populations without cultural adaptation.

Future research should build on the current findings through longitudinal designs to better assess causal relationships between demographic and psychosocial variables and food addiction outcomes. For instance, the associations we observed between younger age, lower educational attainment, and food addiction cannot be interpreted causally due to the cross-sectional design. Incorporating structured diagnostic interviews alongside self-report tools could help improve the precision of prevalence estimates and symptom interpretation. Further investigation into unmeasured psychosocial variables such as mental health conditions, perceived stress, caregiving demands, or employment instability may provide a more comprehensive risk profile.

Given the observed sex-specific symptom patterns, future studies should also explore mechanisms underlying these differences using mixed methods and neurodevelopmental models. The “brain mosaic” framework suggests that symptom expression may result from a combination of biologically and socially influenced traits, regardless of assigned sex84. Exploring this heterogeneity, including underreported symptomatology in males, may improve the gender inclusivity of both diagnostic tools and treatment approaches. Finally, research should continue adapting and evaluating culturally responsive intervention models within Iranian and similar sociocultural settings ideally combining emotion-focused, community-based, and gender-informed components with public health strategies targeting food environments and cultural eating norms.

Conclusion

The prevalence of FA in Iran is 15.6%, with 8.7% for males and 17.2% for females. Weight status, and marital status are significant correlated with FA for both females and males, while sex-specific correlates of FA were also identified (younger age for females whereas lower education for males). Underweight is positively associated with FA for males while negatively associated with FA for females. Sex-specific considerations need to be taken into account when developing interventions of FA in the future.