Abstract

Background

Anxiety disorders are a significant and growing public health concern, impacting individuals鈥� daily lives and professional development. Women exhibit higher rates compared to men. Changes in body mass index (BMI) can affect the mental health of an individual. However, the relationship between BMI and anxiety is unclear. The purpose of this study was to investigate the association between BMI and anxiety disorder.

Methods

BDHS 2022 data were used. Binary logistic regression, restricted cubic spline analysis (RCS), and subgroup analyzes were performed to explore the relationship between BMI and anxiety disorder.

Results

The prevalence of anxiety disorders among ever-married women was approximately twenty one percent. A non-linear, U-shaped relationship between BMI and anxiety disorder was observed, with the lowest risk at a BMI of 22.78 kilograms per square meter. Obesity, as well as underweight, increased the risk of anxiety among the participants, especially in subgroups of participants who were older adults, less educated, lower wealth status, greater age at marriage, and longer cohabitation. For older women, with the lowest risk at a BMI of 25.6 kilograms per square meter, being slightly overweight might serve as a psychological buffer against anxiety. The highest prevalence rate was in formerly married women as well as in the women in the Rangpur division.

Conclusion

This study identified a significant association between BMI and anxiety disorder, revealing a U-shaped relationship where both underweight and obesity were correlated with higher odds of anxiety disorder. Although the results indicate that maintaining a healthy BMI could be associated with a decrease in anxiety levels, the cross-sectional nature of the study prevents establishing a causal relationship. This implies that BMI and anxiety may be correlated, but one does not necessarily cause the other. Future longitudinal studies are needed to explore potential causal mechanisms. The observed association highlights the importance of considering body weight extremes in mental health interventions. These findings underscore the need for integrated public health strategies that address both mental health and nutritional well-being among ever-married women in Bangladesh.

Introduction

Mental health is a fundamental component of overall well-being, encompassing emotional, psychological, and social dimensions. Mental disorders encompass a wide variety of psychological conditions that impact mood, thinking, and behavior, often hindering an individual鈥檚 ability to perform daily activities. Among these, anxiety disorders stand out due to their high prevalence and significant impact. Anxiety disorders are characterized by excessive fear and anxiety, often involving the amygdala, which plays a central role in emotional processing. Globally, anxiety disorders are among the most prevalent mental health issues, affecting millions of individuals across various demographics [1]. According to the World Health Organization (WHO) 2022, approximately one in eight people worldwide experiences a mental disorder, with anxiety and depression ranking as the most common [2]. The Institute for Health Metrics and Evaluation (2019) further reported that anxiety disorders affect 301 million people globally, underscoring the pressing need for effective interventions [3]. Anxiety not only impairs mental health but also influences behavioral responses and physiological well-being, often leading to adverse health outcomes and diminished social functioning [4]. Older adults, in particular, are more vulnerable to anxiety-related health risks, including an increased likelihood of mortality [5]. Emerging research highlights a complex relationship between body mass index (BMI) and anxiety disorders, with studies suggesting both direct and indirect associations. Individuals with obesity or overweight status often experience higher levels of anxiety compared to those with normal weight [6]. A U-shaped association has also been observed, where both underweight and obese individuals are more likely to experience anxiety disorders compared to those with normal BMI [7]. However, conflicting evidence exists, with some studies reporting no significant correlation, particularly in specific populations such as medical students [8]. These inconsistencies highlight the need for further exploration to uncover the nuances of this relationship across different contexts and demographics.

In Bangladesh, anxiety disorders are prevalent across various populations, disproportionately affecting women and adolescents. A nationwide survey by Alam et al. (2023) reported that 4.7% of adults suffer from anxiety disorders, with women exhibiting higher rates compared to men, mirroring global trends [1, 9]. Despite the availability of effective treatments, significant barriers prevent individuals from seeking care. Globally, only one in four people with anxiety disorders receives treatment, with obstacles such as lack of awareness, inadequate mental health services, social stigma, and insufficiently trained healthcare providers posing significant challenges [1, 10]. In Bangladesh, these barriers are compounded by cultural stigma, discrimination, and limited accessibility to mental health services. According to the BDHS (2022), only 12% of women exhibiting symptoms of anxiety or depression sought professional help, while a separate survey conducted by Alam et al. estimated that 94% of individuals with psychiatric disorders in Bangladesh do not receive adequate care [9]. This situation underscores the urgent need for improved mental health infrastructure, and community-based mental health services and support systems could be beneficial in mitigating anxiety disorders [9]. Given the socio-cultural and economic challenges faced by Bangladeshi women, targeted mental health interventions are crucial [11]. Addressing this aligns with Sustainable Development Goal 3 (Good Health and Well-being) by promoting mental health awareness and access to care, as well as SDG 5 (Gender Equality), as women are disproportionately affected by anxiety disorders due to social and economic inequalities [12]. Therefore, The current study aims to investigate the relationship between body mass index (BMI) and anxiety disorders. among ever-married women in Bangladesh, focusing on identifying vulnerable subgroups that require priority attention. By establishing a scientific basis for future prevention and control strategies, this research aims to inform improved treatment and management approaches for anxiety disorders, ultimately enhancing the mental health landscape in Bangladesh.

Methods

Data Source

This study is a cross-sectional analysis utilizing nationally representative data from the Bangladesh Demographic and Health Survey (BDHS) 2022. BDHS follows a two-stage stratified sampling design, selecting Enumeration Areas (EAs) from urban and rural regions based on the 2011 Population Census of Bangladesh. Within each EA, a random sample of households was selected. Data collection occurred in four phases, each lasting approximately four weeks, beginning on June 27, 2022, and concluding on December 12, 2022. The survey covered 30,330 households nationwide, of which 20,220 households were selected for the women鈥檚 long questionnaire. Half of these households were further selected for biomarker measurements, resulting in interviews with 10,053 eligible ever-married women (aged 15鈥�49 years). Although the BDHS dataset includes multiple record files (e.g., IR file, PR file), this study exclusively uses the Individual Record (IR) file. A total of 9,946 cases had complete data on mental health, BMI, and other covariates. As per BDHS 2022 guidelines, sample weights were applied to ensure accurate representation of survey results at national and division levels [13]. After applying sampling weights, the final weighted analytic sample consisted of 9,941 respondents (Fig.听1). The 2022 BDHS included a mental health module that assessed anxiety symptoms using the Generalized Anxiety Disorder-7 (GAD-7) scale, a validated tool with strong reliability and sensitivity [13,14,15]. The scale evaluates symptoms experienced in the two weeks preceding the survey, assigning scores from 0 to 3 based on frequency (BDHS 2022). A total GAD-7 score ranges from 0 to 21, categorized into mild (0鈥�5), moderate (6鈥�14), and severe (15鈥�21) anxiety. To enhance understanding, the Water Glass Pictorial Scale was used for symptom visualization. The dependent variable captures individuals with a GAD-7 score of 6 or higher or those who reported taking prescribed medication for anxiety in the last two weeks, ensuring both symptomatic and treated cases are included in the analysis.

Fig. 1

Flowchart depicting sample selection process in the study

Since this study is based on a nationally representative dataset (BDHS 2022), the sample size was determined by the survey design rather than a pre-specified power calculation. Previous studies on similar topics have used smaller sample sizes, reinforcing that our study provides robust estimates [16, 17].

The BDHS-2022 survey achieved a high response rate, with 99% of eligible women completing the interview. To minimize non-response bias, the survey implemented quality control measures, including multiple follow-up visits, trained interviewers, and electronic data collection with built-in validation checks. Additionally, survey weights were applied to adjust for non-response and ensure the representativeness of the sample. These measures help mitigate potential biases arising from missing responses and enhance the reliability of our findings [13].

Study variables

The study鈥檚 dependent variable was the prevalence of anxiety disorders, while the independent variable was the participants鈥� BMI. The prevalence of anxiety disorder is defined as respondents who either had a GAD-7 score of 6 or higher, indicating moderate to severe anxiety symptoms, or reported taking prescribed medication for anxiety within the two weeks preceding the survey. The independent variable, Body Mass Index (BMI) is measured in kilograms per square meter (kg/m²). This is calculated by dividing a person鈥檚 weight in kilograms by the square of their height in meters. BMI was categorized using the standard cut points, with BMI < 18.5 kg/m², 18.5鈥�24.9 kg/m², 25.0鈥�29.9 kg/m² and $\ge 30.0 \hspace{0.2cm} kg/m^2$ defined as underweight, normal weight, overweight, and obesity, respectively [18]. The selection of other covariates was based on previous literature and the availability of variables in the BDHS-2022 dataset. Since our study focused on ever-married women, we considered marriage-related factors as key confounders in the association between BMI and anxiety disorder. The selection of demographic variables was guided by the mental health chapter of the BDHS 2022 report and relevant literature. This study includes demographic and marriage-related factors such as age, residence, division, education, wealth quintile, terminated pregnancy, child death, current marital status, recent sexual activity, cohabitation duration, age at marriage, and use of modern contraceptive methods. Although some of these variables are directly available in the BDHS 2022 Individual Recode (IR) file, others require recalculation or modification using existing variables in the data set to ensure accurate representation and analysis.

Statistical methods

The data was prepared through a process that included cleaning, creating new variables, and recoding. Subsequently, statistical analysis and data visualization were carried out. A significance level of 0.05 was used for all analyses. Categorical variables were summarized using frequencies and proportions, with group differences assessed via the chi-squared test. Categorical variables were summarized using frequencies and proportions, with group differences assessed using the Chi-square test or Fisher鈥檚 exact test, depending on data distribution. The Chi-square test was used when the expected cell count was $>=$5, ensuring that the assumption of the test was met. In cases where expected cell counts were <5, Fisher鈥檚 exact test was applied to provide a more accurate significant estimate. The association between BMI and anxiety disorder was examined using a binary logistic regression model, while the dose-response relationship was analyzed using a restricted cubic spline (RCS) model.

To examine the association between BMI and anxiety disorder, binary logistic regression analysis was utilized. The logistic regression mathematically models the relationship between a binary dependent variable and independent variables. Specifically, we fitted multiple logistic regression models, starting with a crude (unadjusted) model to directly assess the association between BMI and anxiety without controlling for covariates. We then sequentially introduced covariates: demographic variables (age, residence, division, wealth status, education level), marriage-related variables (age at marriage, cohabitation duration, marital status, recent sexual activity, history of terminated pregnancy, and child death), and then constructed a fully adjusted model including all variables. The final model retained only statistically significant variables. The general logistic regression model is mathematically expressed as:

$$\begin{aligned} \log \left( \frac{P(y=1)}{1 - P(y=1)} \right) = \beta _0 + \beta _1 x_1 + \beta _2 x_2 + \dots + \beta _k x_k \end{aligned}$$

where, $P(y=1)$ represents the probability that the event of interest happens, such as the presence of anxiety. The terms $\beta _0$ and $\beta _i$ are the coefficients estimated from the data, where $\beta _0$ is the intercept (constant term), and $\beta _i$ are the coefficients for the independent variables $x_i$.

Furthermore, to explore non-linear relationships between BMI and anxiety, Restricted Cubic Spline (RCS) analysis was performed: $y = \beta _0 + \sum _{i=1}^{k} \beta _i S_i(\text {BMI})$, where, $S_i (BMI)$ indicates spline transformations of BMI with 3 knots placed at the 10th, 50th, and 90th percentiles, allowing the model to flexibly capture non-linear associations between BMI and anxiety. RCS was chosen over other methods ( like polynomial regression, fractional polynomials, etc.) due to its flexibility and stability in capturing complex nonlinear relationships while minimizing overfitting [19,20,21]. In addition, subgroup analyzes were performed to explore the variations in the association between BMI and anxiety disorder across various demographic and socioeconomic groups.

Results

Profile of the respondents

Table 1 presents the demographic, socioeconomic, and reproductive characteristics of the study participants and their association with anxiety symptoms. Among the 9,941 ever-married women aged 15鈥�49, 21% exhibited moderate to severe anxiety or reported taking prescribed medication for anxiety. BMI was significantly associated with anxiety (p = 0.011), with the highest prevalence among obese (24.85%) and underweight (22.93%) women. Older age ($\ge$35 years) was linked to a higher prevalence (26.04%, $p < 0.001$) compared to younger women (17.98%). While residence (urban vs. rural) did not show a significant difference (p = 0.242), regional variations were evident ($p < 0.001$), with the highest anxiety prevalence in Rangpur (25.91%) and Chattogram (24.33%) and the lowest in Mymensingh (16.75%) and Dhaka (18.18%). Socioeconomic factors played a crucial role, as women with only primary or no education (22.49%, $p < 0.001$) and those in the lower wealth quintile (23.84%, $p < 0.001$) had higher anxiety prevalence compared to those with secondary or higher education (13.63%) and those in the upper wealth category (19.12%). Reproductive history was also significant, with higher anxiety prevalence among women who had experienced a terminated pregnancy (24.94%, $p < 0.001$) or child death (26.46%, $p < 0.001$). Marital status showed a strong association ($p < 0.001$), as divorced, separated, or widowed women reported significantly higher anxiety (41.01%) than currently married women (20.26%). Inactive sexual relationship (25.6%), early marriage ($<15$ years, 24.33%), and longer cohabitation duration ($\ge$10 years, 24.41%) were also significantly linked to higher anxiety prevalence ($p < 0.001$). Additionally, non-use of modern contraception method was associated with higher anxiety (22.91%) compared to users (19.71%, $p < 0.001$), suggesting a complex interplay between reproductive health choices and mental well-being.

Table 1 Background characteristics of participants

These findings highlight the multifaceted nature of anxiety among ever-married women in Bangladesh, emphasizing the need for targeted mental health interventions considering demographic, socioeconomic, and reproductive factors.

Binary logistic regression

To examine the association between BMI and anxiety disorder we employed binary logistic regression models (Table 2). In the unadjusted model (Model 1), obesity was significantly associated with higher odds of anxiety disorder (OR = 1.318, 95% CI: 1.091鈥�1.593), while overweight status showed a marginally significant association (OR = 1.119, 95% CI: 0.995鈥�1.258). The continuous BMI variable also indicated a positive association with anxiety disorder (OR = 1.014, 95% CI: 1.001鈥�1.026). Adjusting for demographic factors in Model 2 yielded similar results, with obesity maintaining a significant association (OR = 1.327, 95% CI: 1.096鈥�1.608), while the effect of overweight status remained marginally significant. In Model 3, which controlled for marriage-related variables, the association between underweight and anxiety disorder became statistically significant (OR = 1.263, 95% CI: 1.047鈥�1.524), while obesity remained a significant predictor (OR = 1.222, 95% CI: 1.009鈥�1.479). However, the continuous BMI variable lost statistical significance (OR = 1.003, 95% CI: 0.99鈥�1.016). In Model 4, which adjusted for all covariates, underweight and obesity remained significantly associated with higher odds of anxiety disorder (OR = 1.22, 95% CI: 1.011鈥�1.472; OR = 1.289, 95% CI: 1.063鈥�1.562, respectively), while the overweight category showed a non-significant association. The final model (Model 5), which retained only significant variables, confirmed the persistent association between obesity (OR = 1.312, 95% CI: 1.084鈥�1.589) and anxiety disorder, while underweight status approached significance (OR = 1.197, 95% CI: 0.993鈥�1.441) (Table 2). These findings suggest a complex, non-linear relationship between BMI and anxiety disorder, where both underweight and obese individuals exhibit elevated risks, even after controlling for key demographic and marital factors. Notably, BMI remained significantly associated with anxiety disorder across all models, confirming its independent association regardless of adjustment for covariates. Since the adjusted models only slightly varied from the crude model, we considered the crude model when comparing associations to maintain consistency. The results underscore the potential role of body weight extremes in influencing mental health outcomes among ever-married women in Bangladesh.

Table 2 Binary logistic regression of BMI and prevalence of anxiety disorder

RCS analysis

The RCS analysis demonstrated a non-linear, U-shaped relationship between BMI and anxiety disorder, with the lowest odds ratio (OR) for anxiety observed at a BMI of 22.78 kg/m².

This pattern was consistent across various subgroups, including individuals aged 45 years or above, those in the lower wealth quantile, individuals with secondary or higher education, those who married at 15 years or older, and those with a cohabitation duration of 10 years or more, with the lowest ORs corresponding to BMI values between 22.4 and 25.6 kg/m². However, this non-linearity was not evident in individuals under 35 years, those in the middle or upper wealth quantiles, individuals with primary or lower education, those married before 15 years of age, or those with shorter cohabitation durations (Fig.听2).

Fig. 2

RCS curve of BMI levels and risk of prevalence of anxiety A in total study population; B in $>=35$ years old; C in primary education or less; D in lower wealth quantile; E in not history of terminated pregnancy; F in no history of child death; G in age at first marriage $>=15$ years; H in cohabitation duration $>=10$ years; I in modern contraception method

Subgroup analysis

The binary logistic regression analysis found significant associations between BMI categories and anxiety across various subgroups (Table 3, Fig.听3).

Table 3 Binary logistic regression analysis of BMI and anxiety in different subgroups

Fig. 3

Forest plot of the association between BMI and anxiety disorder in the different demographic subgroups

Overall, underweight, overweight, and obese women had higher odds of experiencing anxiety compared to those with normal BMI. Age-stratified analysis showed a significant association in women under 35 years but not in those aged 35 or older. Obesity was significantly linked to anxiety in urban areas but not in rural areas. Women with primary or lower education and those in lower and upper wealth quantiles showed significant associations, whereas no such pattern was observed among highly educated or middle-income women.

Obese women who had not experienced a terminated pregnancy or child loss had higher odds of anxiety, while no association was found among those with such experiences. Marital status analysis showed that obese currently married women had significantly higher odds of anxiety and this association was even stronger among divorced, separated, or widowed women. Overweight or obesity increased the odds of anxiety among women experiencing an active sexual relationship, unlike their inactive counterparts. Women who married at 15 years or older showed a significant relationship, whereas those married younger did not. Cohabitation duration of 10 years or more was linked to anxiety in underweight women but not in other BMI categories. Additionally, the use of modern contraceptives was significantly associated with anxiety across all BMI groups, particularly among obese women. These findings highlight that the relationship between BMI and anxiety is influenced by demographic, socioeconomic, and reproductive factors, with some subgroups experiencing stronger associations than others.

Discussion

The present study examined the association between body mass index (BMI) and anxiety disorders among ever-married women in Bangladesh, utilizing data from BDHS-2022. The findings reveal strong association between BMI and anxiety disorder aligning with previous studies conducted in different countries [22, 23]. This study found that 21% of ever-married women in Bangladesh suffer from anxiety disorders (Table 1), which is consistent with the nationwide survey estimate of 20.4% reported in BDHS-2022 [13]. The slight variation may stem from differences in sample size, as our analysis was restricted to individuals with complete BMI data. Nonetheless, the high prevalence of anxiety highlights the significant burden of mental health issues among Bangladeshi women, underscoring the need for targeted mental health interventions. Our study observed a non-linear association between body mass index (BMI) and anxiety disorders among ever-married women in Bangladesh, utilizing data from BDHS-2022. Similar results were found in a study of depression, where there was a U-shaped relationship between BMI and depression, where both obesity and underweight increased the risk of depression among the participants [16]. We detected a relationship between different categories of BMI and the prevalence of anxiety using binary logistic regression, presented in Table 2, with both underweight and obese women exhibiting higher odds of anxiety compared to those with normal BMI. This aligns with global literature indicating that obesity and being overweight are associated with increased risks of mood, anxiety, and personality disorders [24]. While underweight individuals also experience certain mental health disorders, these associations are generally less pronounced compared to those observed in obesity [24]. In Bangladesh, there exists a dual burden of malnutrition, with both underweight and overweight issues being prevalent among women [25]. In this context, underweight women often experience economic hardships, malnutrition, and social insecurities, all of which can contribute to heightened anxiety levels [26]. Conversely, overweight and obese women frequently face weight stigma, societal pressure, and health concerns, which may further exacerbate psychological distress [27]. Thus, both malnutrition and obesity can significantly elevate stress and anxiety, given their profound impact on physical health and social well-being [28, 29]. In the results of the RCS analysis, it was evident that the relationship between BMI and the prevalence of anxiety disorder was a U-shaped association among the ever-married women analysis (Fig.听2). The lowest OR for anxiety was observed at a BMI of 22.8 kg/m², which corresponds to the normal BMI range. This pattern was consistent across various subgroups. RCS analysis of different age subgroups showed a U-shaped relationship between BMI and the risk of anxiety disorder in older women ($>= 35$ years) with the lowest odds ratio for the BMI level of 25.6 kg/m². The U-shaped association was also observed in the subgroups of women with lower wealth status, secondary or higher education, older age at first marriage ($>=15$ years), and a longer duration of cohabitation ($>=10$ years), with the lowest OR observed at BMI 22.7 kg/m², 22.7 kg/m², 22.4 kg/m², and 25 kg/m² respectively. Some studies have reported a positive association between BMI and mental disorders. For instance, a study conducted in China found that higher BMI was linked to poor mental health outcomes, including anxiety and depression, among nurses during the COVID-19 pandemic, with the association being particularly pronounced in those classified as overweight or obese [17]. Similarly, in our study, certain subgroups exhibited a positive association between BMI and anxiety, where the risk of anxiety prevalence increased with rising BMI (Table 3). A statistically significant OR was found for both overweight and obese women in subgroups characterized by younger age, lower education levels, upper wealth quantile, and the women who were having active sexual relationships. These findings indicate that there exists a strong association between BMI and anxiety disorder, although the pattern of the relationship is not uniform across all the subgroups of populations but rather varies based on sociodemographic and behavioral factors. A finding of this study was the unusual scenario observed among women aged 35 years or older and those with at least 10 years of cohabitation: U-shaped association. Unlike most subgroups, where the lowest odds ratios fell within the normal BMI range, these values classify as overweight according to standard BMI categories. This phenomenon could be explained by the "jolly fat" hypothesis, which suggests that higher body weight may be associated with greater psychological well-being [30]. Several studies on middle-aged and older adults have found a negative association between BMI and mental disorders such as depression, indicating that higher BMI might act as a protective factor against psychological distress. That is, psychological stress caused by rapid weight loss, such as diet control, may also increase the risk of mental disorders [30, 31]. However, the 鈥渏olly fat鈥� hypothesis for older women remains a theoretical perspective rather than a definitive explanation, and findings on this relationship have been inconsistent. Further research is needed to explore the psychological and physiological mechanisms underlying this association. The study emphasizes the importance of developing targeted mental health policies in Bangladesh to meet Sustainable Development Goal 3 (SDG 3). This includes expanding mental health care services, reducing stigma, and ensuring that women, especially in underserved communities, have equal access to mental health support, which also aligns with Sustainable Development Goal 5 (SDG 5) [12].

Conclusion

This study examined the association between BMI and anxiety disorder among ever-married women aged 15 to 49 years in Bangladesh using BDHS-2022 data. The findings indicate a U-shaped relationship, where both underweight and obesity are significantly associated with higher odds of anxiety disorder. The lowest risk was observed at a BMI of 22.78 kg/m², with a higher threshold for older women. Subgroup analyses revealed that women with lower education, lower wealth status, early marriage, longer cohabitation duration, and those not using modern contraceptive methods had a higher prevalence of anxiety. Moreover, women who were no longer in active sexual relationships with their legal partner exhibited significantly higher anxiety disorder rates. Disruptive life experiences, such as having experienced at least one child death or a terminated pregnancy, were also associated with an increased risk of anxiety disorders. Additionally, separated, divorced, or widowed women had substantially higher anxiety prevalence compared to currently married women. Notably, women in Rangpur Division were more likely to experience anxiety disorders than those in other regions.

Limitations.

This study has several limitations. First, the cross-sectional design prevents us from establishing causal relationships between BMI and anxiety disorder. Second, while we adjusted for multiple confounders, the possibility of residual confounding remains, as certain factors such as lifetime psychiatric history and chronic illness were not available in the dataset. Additionally, the dataset did not include information on family history of mental health disorders, past traumatic experiences, or other key psychological factors strongly associated with anxiety. The inclusion of such variables could have provided a more comprehensive understanding and improved the study findings. Third, anxiety disorder was assessed using the GAD-7 questionnaire, a self-reported tool, which may introduce recall bias and social desirability bias. Although GAD-7 is a validated screening instrument, self-reported responses are inherently subjective. Finally, while BDHS 2022 followed a robust sampling strategy, non-response bias cannot be entirely ruled out. However, the high response rate (99%) and the application of sampling weights help ensure the representativeness of the findings at the national and division levels.