Association between self-reported sleep quality and body mass index among Jimma University academic staff: structural equation modeling

Abstract

Background

In 2022, over 890 million adults globally were affected by obesity. That year, around 16% of adults aged 18 and above were classified as obese. The global obesity rate more than doubled from 1990 to 2022. By 2035, the World Obesity Atlas predicts that over 1 billion people worldwide will be considered obese, which translates into 1 in 5 women and 1 in 7 men will be affected globally. Sleep quality has declined simultaneously with the increased prevalence of overweight and obesity, suggesting a potential link. However, inadequate published articles shows that poor sleep quality is a risk factor for obesity in young and older adults in Ethiopia.

Methods

An institution-based cross-sectional study of Jimma University academic staff was conducted. A total of 427 academic staff members participated in the study. A two-stage cluster sampling procedure was used to select study participants from their departments. Height and weight measurements were taken by trained data collectors. A reliable self-administered questionnaire was used to assess sleep quality. Analysis was performed using Stata version 13.1. Structural equation modeling using the maximum likelihood estimation method was used to analyze the data.

Results

A total of 32.3% (95% CI: 28.0, 36.9) of the participants had poor sleep quality. The mean (± SD) BMI of the respondents was 22.7 (± 3.1) kg/m². The study indicated that 23.1% of the academicians at Jimma University had a BMI greater than 25 kg/m². The overall prevalence of depression, anxiety, and stress was 25.5%, 44.7%, and 16.62%, respectively. Poor sleep quality appeared to have a significant inverse and indirect association with BMI (β = -0.08/P = 0.042), which was mediated through depression and obesogenic dietary behavior.

Conclusion

The present study revealed that poor sleep quality is inversely associated with BMI among Jimma University academic staff. It is highlighted in the study that there is a significant interplay between depression, obesogenic eating behavior, and poor sleep quality in influencing healthy weight. By focusing on lifestyle modifications, behavioral therapies, and restyling the working environment, individuals may experience improvements in both their sleep patterns and weight management outcomes. Future research should continue to explore the complex relationships between sleep quality, lifestyle factors, and BMI to refine these interventions further.

Supplementary Information

The online version contains supplementary material available at 10.1186/s40795-025-01154-7.

Introduction

Obesity has become a major global public health issue, with over 2.5 billion adults classified as overweight and more than 890 million as obese, according to the World Health Organization (WHO) 2024 [1]. The prevalence of obesity is expected to increase significantly, with the World Obesity Atlas predicting that by 2030, one billion people worldwide will be living with obesity, affecting 1 in 5 women and 1 in 7 men globally [2]. Although obesity rates in sub-Saharan Africa are still lower than those in high-income countries, they are rising rapidly, with regional prevalence rates from 5.6% in Madagascar to 27.7% in Swaziland [3, 4]. In Ethiopia, the prevalence of overweight and obesity is 12.1% and 2.8%, respectively, with higher rates seen in urban areas compared to rural regions [5].

The rise in obesity is attributed to various factors, including sedentary lifestyles, excessive alcohol consumption, and unhealthy dietary habits [15]. Additionally, socioeconomic factors such as higher education and wealth have been linked to increased obesity risk in low- and middle-income countries [16–20]. Concurrently, sleep quality has declined globally, with modern lifestyles, technological advancements, and increased screen time contributing to chronic sleep deprivation [6, 7]. Sleep deprivation has been identified as an independent risk factor for obesity, as it disrupts metabolic and hormonal regulation, leading to increased appetite, unhealthy dietary choices, and reduced physical activity [8–12]. Studies have shown that poor sleep quality is associated with higher body mass index (BMI) and obesity, particularly in developed countries [13–17]. However, evidence from African countries, including Ethiopia, remains scarce.

During adulthood, the need for sleep becomes reduced compared to healthy children and adolescents due to lifestyle choices, including family or work commitments, and psychological or physical problems. Voluntary sleep reduction has become increasingly common in today’s modern world [6]. The current technological advances have resulted in a population group, mainly individuals with a sedentary behavior, which is operational 24/7 and may have altered leisure time behaviors with individuals spending their time watching television, video gaming, using mobile telephones, and internet surfing. Use of these media may result in chronic sleep deprivation through delayed bedtime [7].

In recent years, sleep deprivation has gained increasing attention as an independent risk factor for overweight and obesity in adolescents and adults [8, 9]. Many studies in developed countries have shown that sleep deprivation could increase energy intake by giving people more time to eat and promoting people to choose less healthy diets as well as by increasing hunger and decreasing energy expenditure by decreasing physical activity and lowering body temperature [10–12].

Morselli and his colleagues hypothesized that there are several ways that sleep deprivation, both in quality and quantity, might lead to weight gain, either by increasing how much food people eat or decreasing the energy that they burn [13]. Sleep loss is associated with metabolic and hormonal alterations, which can be mainly explained by decreased glucose tolerance and alterations in appetite-regulating hormones since sleep is an important modulator of neuroendocrine function and glucose metabolism [14, 15].

Jennings and colleagues conducted a study on 210 participants (57% men) and discovered that an increase in body mass index (BMI) was significantly associated with poor sleep quality [16]. A meta-analysis by Fatima and her colleagues revealed that sleep deprivation was associated with being overweight or obese (OR 1.46) [17].

Studies investigating the association between sleep deprivation and obesity have been widely conducted in developed nations, and there is very little evidence from African nations. However, a study performed in Soweto, South Africa, revealed a negative relationship between nighttime sleep duration and BMI in both sexes. Nevertheless, a lower BMI and WC were observed in males who slept during the day than in those who did not [18].

Several limitations may impede the generalizability and consistency of the identified findings. First, the underlying individual variations that affect both sleep quality and BMI were not assessed. For instance, income was not considered, nor was obesogenic dietary behavior or behavioral factors that might be driven by psychological factors, such as stress, anxiety, and depression.

Although associations between poor sleep quality and increased BMI have been reported in many studies, few studies have tested all the model components simultaneously. As both sleep quality and potential mediators, i.e., depression and obesogenic dietary behavior, are latent constructs, a model that takes into account the latent structure should be used. Therefore, this study aimed to examine the pathways through which poor sleep quality is linked to an increase in body mass index after controlling for other factors. The magnitude of both variables, poor sleep quality and higher BMI, was also determined. The findings of this study can benefit policymakers in designing appropriate interventions, especially for the population group that follows a sedentary lifestyle, in order to tackle the emerging problem of overweight/obesity and its associated factors among the study population. Generally, the study will provide preliminary information for further research in the study area of overweight/obesity and sleep quality. The findings are expected to improve the planning of interventions for preventing the worsening of obesity among the population.

Methods and materials

Study design and period

An institution-based cross-sectional study was conducted at Jimma University from February 01 to March 31, 2019. There are four campuses at the university (main campus, technology campus, college of business and economics, and agricultural campus), with 8,000 staff, 1661 of which are teaching staff at six colleges and two institutes.

Sample size determination

Sample size calculations were conducted for each of our three objectives, and the largest sample size was considered for the study.

Objective 1 To determine the prevalence of obesity among Jimma University academic staff

The sample size was determined by using the Epi Info™ 7 by considering the following assumptions: 11.3% overall prevalence of overweight/obesity in Bahir-Dar city [19], 5% margin of error, 5% nonresponse rate, 1661 source population, and a final sample size of 148.

Objective 2 To determine the magnitude of sleep quality among Jimma University staff

The sample size was determined by using an Epi Info™ 7 by considering the following assumptions: 50% overall magnitude of sleep deprivation, 5% margin of error, 5% nonresponse rate, 1661 source population and a final sample size of 327.

Objective 3 This study aimed to predict the association between sleep quality and BMI mediated through dietary pattern and physical exercise.

A sample size of 316 subjects was used to detect a correlation of at least 0.16 between sleep deprivation and BMI with a 95% confidence level, a power of 80%, and an anticipated 5% nonresponse rate. Therefore, the minimum required sample size for this study was 316, with a 5% nonresponse rate. The calculation formula is based on a two-tailed test. The sample size for correlation analysis was generated with G-power software.

The largest number among all three objectives was used, and the final sample size was 327. Clustering by department was performed to select study participants, and thus, a design effect of 1.5 was used to yield a total sample size of 491.

Sampling technique

A two-stage cluster sampling procedure was employed to select study participants from among the JU academic staff. First, 30% of the study departments were selected from among the total departments on the campus. Then, each academic staff member from each selected department was included in the study. Computer-generated random numbers were used to select the department based on lists of departments.

Inclusion criteria

All Jimma University teachers on campus during the data collection period were eligible for the study.

Exclusion criteria

• Pregnant women

• Subjects taking sleep or psychotherapy medications during the study that could directly affect sleep patterns were excluded.

Data collection method

All the data, except the anthropometric measurement, were collected using a self-administered questionnaire.

Measurements tools

Sleep quality

Sleep quality was assessed using the Pittsburgh Sleep Quality Index (PSQI) developed by Buysse et al. [20]. The PSQI consists of 7 elements, including perceived sleep quality, sleep latency, sleep duration, sleep efficiency, sleep disturbances, use of sleeping medication, and daytime dysfunction. There are 19 items that are used to generate 7 component scores, which are then added together to determine a global score that ranges between 0 and 21, where higher scores reflect worse sleep quality. This tool has been validated to be used in the Ethiopian context [21].

Eating behavior

Eating behavior data were collected using the Eating Behavior Pattern Questionnaire (EBPQ) [22]. The EEBPQ is a validated questionnaire designed to measure nine dimensions of human eating behavior. This questionnaire consists of 51 self-reported items on healthy and unhealthy eating behaviors. Every item was rated on a 5-point Likert scale ranging from strongly disagree to strongly agree. Six eating behavior patterns were assessed by the questionnaire including low fat eating (11 total items), snacking and convenience (10 total items), emotional eating (8 total items), planning (6 total items), meal skipping (7 total items), and cultural/lifestyle behavior (9 total items). Nighttime eating and stimulant use were assessed via direct questionnaire items. Additional questions were added that have been used to assess intake behavior, intake distribution, and intake timing.

Psychological factors

Psychological factors, i.e., depression, anxiety, and stress, were measured using Lovibond and Lovibond’s short version of the Depression Anxiety and Stress Scale (DASS-21) [23]. DASS-21 is a validated and reliable instrument with 21 items in three domains. Each domain comprises seven items assessing symptoms of depression, anxiety, and stress. Participants will report indicating the presence of symptoms in each domain over the past week, scoring from 0 (did not apply at all) to 3 (applied most of the time). Scores from each dimension will be summed. Then, the final score was multiplied by 2 and then categorized according to the DASS manual as normal, mild, moderate, severe, and extremely severe.

Physical activity

Core items from the WHO STEPwise approach to chronic disease risk factor surveillance (STEPS) instrument were used to assess the physical activity level of the participants [24].

Anthropometry

Height was measured using a height measuring board to the nearest 0.1 cm while the subjects were on the Frankfurt plane and had no shoes on. Weight was measured twice to the nearest 0.1 kg using a digital weight scale while subjects had no shoes and minimal clothing. Body mass index was then calculated by dividing weight (in kilograms) by the square of height (in meters). BMI values were categorized based on the WHO-recommended cutoff for underweight, normal, overweight, and obese individuals.

Covariates

Potential covariates were identified a priori. These included age, sex, academic rank, education level, marital status, additional income, perceived workload, and medical disorder history, i.e., hypertension, diabetes mellitus, and obstructive sleep apnea (OSA).

Data analysis

To provide a sound explanation of how participants’ sleep quality could predict BMI through their dietary pattern and/or physical activity, structural equation modeling (SEM) was conducted. Accordingly, it is hypothesized that poor sleeping habits would result in weight gain and/or an increase in BMI that would be mediated through 1) participants’ obesogenic dietary behavior driven by depression and 2) subjects’ physical inactivity, which would ultimately result in an increase in BMI.

SEM with maximum likelihood estimation was used to model the above hypothesized relationships. In the measurement models, factor analysis was employed to estimate the latent variables, dietary pattern, sleep quality, and perceived depression status from their observed construct variables. In the structural models, estimations of both unstandardized and standardized estimates of the direct and indirect effects of sleep quality on BMI through the hypothesized pathways are made. The need for further model adjustment for relevant variables was checked by including important covariates. The reliability of the measurement scales and the relative importance of each construct variable in a scale were evaluated using Cronbach’s alpha coefficient. Model fit was evaluated by utilizing the chi-square statistic, standardized root mean square residual (SRMR), Tucker‒Lewis index (TLI), comparative fit index (CFI), and root mean square error of approximation (RMSEA). As proposed by Hu and Bentler [23], TLI and CFI values ≥ 0.90, SRMR values ≤ 0.08, and RMSEA values ≤ 0.06 were defined to indicate an adequate fit for the data.

Results

The study included 427 participants, with 79.6% being male and a mean age of 31.48 (± 1.24) years for males and 27.34 (± 2.05) years for females. A total of 32.3% of participants were classified as having poor sleep quality based on a global Pittsburgh Sleep Quality Index (PSQI) score of ≤ 5. Poor sleep quality was more prevalent among males (35.6%) compared to females (19.5%), with a statistically significant difference (p = 0.004). Participants aged 45 years and older had a higher prevalence of poor sleep quality (41.2%) compared to younger age groups (33.8% for ages 25–34 and 35.8% for ages 35–44). The difference was statistically significant (p = 0.026 for > 45 years and p = 0.005 for 25–34 years).

Only 38.6% of participants met the WHO guidelines for physical activity. However, there was no significant association between physical activity levels and sleep quality (p = 0.24). The overall prevalence of depression was 25.5%, with 16.6% experiencing moderate depression and 8.9% experiencing severe depression. Poor sleep quality was significantly associated with higher levels of depression (p < 0.001). The prevalence of anxiety was 44.7%, with 27.9% experiencing moderate anxiety and 16.9% experiencing severe anxiety. Poor sleep quality was significantly associated with higher levels of anxiety (p < 0.001). he prevalence of stress was 16.62%, with 11.2% experiencing moderate stress and 5.4% experiencing severe stress. Poor sleep quality was significantly associated with higher levels of stress (p < 0.001).

Participants with excessive workload had a lower prevalence of poor sleep quality (22.7%) compared to those with high (33.1%) or comfortable workloads (32.9%). The association between workload and sleep quality was statistically significant (p = 0.002 for high workload and p = 0.004 for comfortable workload). Single participants had a higher prevalence of poor sleep quality (33.6%) compared to married participants (31.3%). However, the difference was not statistically significant (p = 0.38). Divorced participants had no reported poor sleep quality, but the sample size was small (n = 3) (see Supplementary Table 1, Additional File 1).

Table 1.

Summary of direct and indirect SEM pathways linking sleep quality to BMI before and after covariate adjustment among Jimma University academic staff. Standardized β coefficients, 95% confidence intervals (CIs), and p-values are reported for both unadjusted and adjusted models. The adjusted model includes covariates: age, sex, income, and diabetes history. Mediation types indicate whether pathways were fully or partially mediated. Notable changes in effect sizes highlight covariate influence on key psychological and behavioral mechanisms

Pathway	Unadjusted β (95% CI)	p-value	Adjusted β (95% CI)	p-value	Interpretation	Mediation Type	Notes
Sleep → Depression	0.69 [0.61, 0.78]	< 0.001	0.69 [0.61, 0.78]	< 0.001	Poor sleep increases depression consistently	Direct	Stable across adjustments
Depression → Diet	0.43 [0.21, 0.65]	< 0.001	0.58 [0.35, 0.81]	< 0.001	Depression strongly increases unhealthy diet	Partial	Strengthened after income adjustment
Diet → BMI	−0.18 [–0.32, –0.05	0.007	−0.22 [–0.37, –0.06]	0.005	Unhealthy diet is associated with higher BMI	Direct	Slightly strengthened with diabetes included
Sleep → Diet	−0.07 [–0.29, 0.15]	0.543	−0.05 [–0.28, 0.18]	0.686	No direct path from sleep to diet	Fully mediated	Non-significant in all models
Sleep → BMI (indirect via Depression → Diet)	−0.06 [–0.14, 0.02]	0.128	−0.08 [–0.16, –0.01]	0.042	Sleep influences BMI indirectly	Fully mediated	Significant only after adjustment
Sleep → Physical Activity	0.10 [0.01, 0.20]	0.032	0.11 [–0.06, 0.28]	0.192	Lost significance after age adjustment	N/A	Age likely confounded the relationship
Physical Activity → BMI	0.06 [–0.07, 0.19]	0.367	0.04 [–0.11, 0.20]	0.584	No significant effect	N/A	Remained nonsignificant

Associations between sleep quality and BMI

The main aim of this study was to predict the path through which poor sleep quality is correlated with BMI among Jimma University academic staff. The mechanism of action for the proposed associations was through physical inactivity, depression, and obesogenic dietary behavior. Sleep quality, dietary pattern and depression were the latent constructs, whereas physical activity level and BMI were observed to be variables.

Structural equation model

Structural Equation Modeling (SEM) was employed to assess both direct and indirect pathways linking sleep quality to BMI, with depression and dietary behavior as potential mediators. SEM provides the flexibility to specify models based on theoretical frameworks and prior research findings. Since the study variables consists of both latent and observed variables, SEM is a robust statistical technique used to examine complex relationships among both variables, while accounting for measurement error.

The final adjusted model incorporated covariates like age, sex, income, and diabetes history based on theoretical and empirical evidence of their potential confounding effects on the sleep-depression-diet-BMI pathway. Age and sex were included due to their well-documented associations with sleep quality, mental health, and metabolic outcomes, while income was adjusted for as a proxy for socioeconomic status, which influences access to healthier diets and recreational physical activity. Diabetes history was included as a clinical confounder, given its bidirectional relationship with both depression and BMI. These covariates were selected a priori to isolate the specific mechanistic pathways between sleep, depression, and BMI, ensuring that the observed associations were not attributable to demographic or health-related heterogeneity. Sensitivity analyses confirmed that covariate adjustment strengthened key pathways (e.g., depression → diet) and revealed indirect effects (e.g., sleep → BMI) otherwise masked in unadjusted models, underscoring the importance of their inclusion for robust causal inference.

In both unadjusted and adjusted models, poor sleep quality was strongly associated with increased depression symptoms (β = 0.69, p < 0.001, 95% CI: 0.61—0.78), indicating a stable and direct effect. Depression, in turn, significantly predicted unhealthy dietary behaviors, and this effect strengthened after adjusting for covariates (β increased from 0.43 to 0.58, p < 0.001), suggesting that income may amplify stress-induced eating patterns. Unhealthy dietary behavior was negatively associated with BMI (β = −0.22, p = 0.005, 95% CI: −0.37 to −0.06), indicating that individuals with poorer diets had higher BMIs (Table 1).

Importantly, the direct path from sleep quality to dietary behavior was not significant in either model (β = −0.05, p = 0.686), confirming full mediation through depression. Likewise, no direct effect was found between sleep quality and BMI. However, an indirect pathway from sleep to BMI emerged as statistically significant only in the adjusted model (β = −0.08, p = 0.042, 95% CI: −0.16 to −0.01), mediated sequentially by depression and dietary behavior (Figure 1). This indirect association suggests a suppressed relationship that was revealed only after accounting for key covariates.

Fig. 1.

Structural Equation Model illustrating the adjusted pathways between sleep quality, depression, dietary behavior, and body mass index (BMI). Poor sleep quality was significantly associated with higher depression levels (β = 0.69, p < 0.001), which in turn predicted poorer dietary behavior (β = 0.58, p < 0.001). Poor dietary behavior was associated with higher BMI (β = −0.22, p = 0.005), indicating an indirect pathway from sleep to BMI through depression and diet. The direct path from sleep quality to dietary behavior was not significant (β = −0.05, p = 0.686), underscoring depression as a key mediator. Paths involving physical activity were not statistically significant. These findings suggest that sleep quality influences BMI primarily through emotional and behavioral mechanisms, though caution is warranted in interpretation

Paths involving physical activity were non-significant after adjustment, indicating that behavioral links between sleep and BMI in this population operate primarily through emotional and dietary mechanisms rather than energy expenditure. The final model achieved acceptable fit (χ²(75) = 108.53, p = 0.007; RMSEA = 0.06; CFI = 0.93; TLI = 0.92) Fig. 1.

Sensitivity analysis: impact of covariate adjustment

Adjustment for key covariates, including age, sex, income, and diabetes history, meaningfully refined the structural pathways in our model. The association between poor sleep quality and depression remained robust and unchanged (β = 0.69, p < 0.001, 95% CI [0.61, 0.78]), suggesting that this relationship is stable across demographic and clinical contexts. In contrast, the path from depression to unhealthy dietary behavior strengthened substantially after adjustment, increasing from β = 0.43 (95% CI [0.21, 0.65]) to β = 0.58 (p < 0.001, 95% CI [0.35, 0.81]), indicating that income may intensify the behavioral impact of depressive symptoms by enabling greater access to high-calorie foods. Similarly, the dietary behavior to BMI pathway became stronger (from β = −0.18, 95% CI [−0.32, −0.05] to β = −0.22, 95% CI [−0.37, −0.06], p = 0.005) when controlling for diabetes, reflecting the influence of clinical status on weight-related outcomes. Meanwhile, previously weak associations between sleep quality, physical activity, and BMI were rendered non-significant after accounting for age and sex, implying these links are primarily demographic rather than behavioral. Most notably, a significant indirect effect of sleep quality on BMI (β = −0.08, p = 0.042, 95% CI [−0.16, −0.01]) emerged only in the adjusted model, emphasizing the importance of covariate control in uncovering subtle mediated pathways. Collectively, these results show that while some associations are consistent across contexts, others are strongly moderated by socioeconomic and health factors, which must be accounted for to accurately interpret behavioral mechanisms.

Psychometric properties of tools

The reliability of the tool was checked based on the assumption that for every latent variable, the Cronbach’s alpha value must be equal to or greater than 0.7 for a tool to be valid. Cronbach’s alpha values for sleep quality assessment, dietary assessment, and psychological assessment were 0.77, 0.94, and 0.78, respectively. All indices are greater than 0.7. Therefore, the validity of the research model is acceptable. Factor loadings for Sleep (λ = 0.76–0.82) and Depression4 (λ = 0.79–0.88) indicated excellent construct validity, while Dietary Behavior showed acceptable loadings (λ = 0.68–0.72). All scales demonstrated reliable measurement (α > 0.70), supporting their use in structural analyses.

Discussion

This study indicated that 23.1% of academicians at Jimma University had a BMI greater than 25 kg/m² and were thus overweight or obese. This percentage was much greater than that reported in the Ethiopian Demographic Health Survey (EDHS) [5], which reported 14.9% and 11.3%, respectively, in a study performed in Bahir-Dar [19]. This finding is lower than that of a study in Tanzania, in which 43.3% of participants had a BMI greater than 25 [26]. This discrepancy might be due to differences in sociodemographic characteristics and dietary behaviors.

The findings from this study offer nuanced support for existing theories regarding the association between sleep quality and BMI, while also presenting refinements that underscore the value of using more sophisticated analytic techniques like SEM. Traditional models often posit a direct relationship between poor sleep and increased BMI, grounded in evidence that sleep deprivation disrupts hormonal regulation (e.g., ghrelin and leptin), increases appetite, and reduces energy expenditure [27]. However, the current study challenges this simplified direct effect model by demonstrating that sleep quality does not directly influence BMI, but instead acts through indirect pathways involving psychological and behavioral mediators, specifically depression and dietary behavior.

The SEM results strengthen an emerging theoretical framework that considers mental health as a critical intermediary between sleep and weight outcomes. Poor sleep was found to significantly increase depression, which then led to unhealthy eating patterns, and these patterns directly influenced BMI. This multi-step mediation model supports recent integrative perspectives suggesting that psychological distress is a key mechanism linking sleep disturbances to metabolic and weight outcomes [28]. Moreover, the indirect effect of depression on BMI also alludes to a dual pathway: one through altering dietary behavior and another potentially through biological mechanisms such as altered metabolism or appetite suppression, observed in subtypes of depression.

This study revealed that the prevalence of poor sleep quality among the study participants was 32.3%, which is consistent with findings in China (39.4%) and the USA (40%) [29, 30]. In relation to the significant sex difference in poor sleep quality, a greater proportion of poor sleep quality was observed among males (35.6%), which contradicts the findings of studies conducted in South Africa [31] and China [32]. This might be because more male academics than female academics participated in this study.

The findings of the present studies are in line with those of a prospective study conducted in Dutch [33] and mid-life US adults [34], where depression plays a mediation role. It is possible that sleep deprivation has the potential to alter one’s ability to address depression daily and involuntarily forces individuals to engage in unhealthy dietary intake behaviors that can increase their chances of becoming obese. Those in negative emotional states have been shown to favor the consumption of rewarding foods high in sugar and/or fat, whereas intake during happy states favors less edible dried fruits [35].

It is important to reiterate that the observed inverse association between poor sleep quality and BMI in our study should not be interpreted as a protective physiological effect. Rather, it likely reflects underlying economic hardship, food insecurity, and energy constraints common in low-income settings. The non-linear trend shown in Fig. 2 further supports this interpretation by suggesting a U-shaped relationship, where both under- and overweight individuals may coexist among individuals experiencing poor sleep quality (Fig. 2). But the total effect of poor sleep quality on BMI through those mediators appears to be inverse. Although there are few such studies, there are earlier findings that have revealed an inverse association between BMI and poor sleep quality. The findings of the present study are consistent with those of two studies performed on adults in a university setting. Soares and his colleague’s studies among university students in different age groups [36] and a gender-specific study conducted in China that targeted females rather than males from the general population [37] were among the few. A study performed in Soweto, South Africa, also revealed a negative relationship between sleep deprivation and BMI in both sexes [18].

Fig. 2.

Distribution of BMI across global PSQI scores among study participants. This figure illustrates a non-linear association between BMI and PSQI scores, suggesting possible U-shaped dynamics influenced by socioeconomic context. The income disparity suggests socioeconomic status moderates the sleep-BMI relationship, as food access may override appetite dysregulation from poor sleep. While poor sleep has been associated with increased appetite in other contexts, limited financial resources may constrain food access and caloric intake in this setting. These findings should be interpreted with caution, as unmeasured factors such as food insecurity, cultural dietary practices, and health conditions may also contribute to the observed association

The current finding was also inconsistent with many of the findings. A study conducted at Pittsburgh University, Pennsylvania, revealed a significant relationship between poor sleep quality and high BMI [16]. These results may also indicate that difficulties initiating or maintaining poor sleep and short sleep duration may not be correlated, and consequently, their associations with weight/obesity may be different.

One possible explanation for the unexpected inverse relationship between poor sleep quality and BMI may lie in the economic constraints experienced by the study population. In low-income contexts such as Ethiopia, individuals with lower socioeconomic status might face limited access to food regardless of their sleep patterns. Further exploration of the relationship between global PSQI scores and BMI (see Fig. 2 below) revealed a non-linear pattern, with both lower and higher BMIs observed among sleep-deprived individuals. Notably, participants with lower BMIs reported a mean monthly income of 9,095 ETB, whereas those with higher BMIs had a mean income of 14,992 ETB, suggesting that economic status may significantly influence food access and, consequently, BMI. In this setting, individuals with limited income may be unable to consume sufficient calories even if poor sleep heightens appetite and food cravings, as has been documented in other populations. However, these findings should be interpreted with caution. The complex interplay between sleep, appetite regulation, and energy intake may be moderated by context-specific factors such as cultural eating patterns, food security, and underlying health conditions.

Moreover, the study found that only 14.86% of participants reported nighttime eating due to sleeplessness, and the majority (74.47%) consumed meals only three times a day. This indicates that frequent snacking and irregular eating behaviors, which are typical obesogenic behaviors, may not be prevalent in this population due to limited food access. In such settings, poor sleep quality may not translate into increased caloric intake because individuals simply do not have the resources to consume more food, even if they feel hungry.

Dietary habits may also play a role in the observed inverse association. In many low-income countries, including Ethiopia, traditional diets are often plant-based and low in fat, which may contribute to lower BMIs even in the presence of poor sleep quality [38]. The study assessed dietary behavior using the Eating Behavior Pattern Questionnaire (EBPQ), which includes dimensions such as low-fat eating, meal skipping, and cultural/lifestyle behaviors. Cultural norms around meal timing and food choices may mitigate the impact of poor sleep on BMI. For example, if cultural practices emphasize regular meal times and discourage snacking, this could reduce the likelihood of weight gain even among individuals with poor sleep quality.

Additionally, individuals may rely on stimulants like caffeine or nicotine to cope with fatigue resulting from poor sleep. These substances can suppress appetite and contribute to weight loss, even in sedentary individuals [39].

Notably, the study's findings also contradict earlier models that emphasized physical activity as a mediator between sleep and BMI, as no significant associations were found between sleep quality and physical activity or between activity levels and BMI. This challenges assumptions in models like the energy balance theory, which posit that physical activity is a primary behavioral mediator of weight change. However, it is worth considering that a sedentary lifestyle, combined with poor sleep, could lead to irregular eating patterns, such as skipping meals or eating at inconsistent times [40, 41]. In the presence of limited food access, this could result in insufficient caloric intake and lower BMI. The study participants had a relatively low level of physical activity, with only 38.64% meeting WHO guidelines for physical exercise. This sedentary behavior, coupled with poor sleep, might not lead to weight gain if food access is limited, as the energy expenditure is already low. This is in line with the findings for the population group with a sedentary lifestyle, where such lifestyles have been found to correlate with higher BMI due to reduced energy expenditure. On the other hand, individuals who experience food access limitation may not consume enough calories to maintain a higher weight, despite their sedentary lifestyle [42].

The study highlights a bidirectional relationship between poor sleep quality and psychological distress. Poor sleep can lead to increased depression, anxiety, and stress, while these psychological conditions can further disrupt sleep, creating a vicious cycle. This cycle can negatively impact dietary choices and physical activity levels, ultimately affecting BMI [43]. The SEM results suggest that interventions targeting sleep quality could have a cascading positive effect on mental health, dietary behaviors, and weight management.

To the best of the authors’ knowledge, this is the first report attempting to employ a structural equation model to understand the link between poor sleep quality and BMI among adults living in Ethiopia. These findings contribute to the growing body of scientific evidence on the impact of sleep deprivation on healthy adult metabolic function by revealing the direct, indirect and total effects of poor sleep quality on weight status.

The findings of the present study should be interpreted considering the most recent study limitations. First, given the cross-sectional nature of our study, it is difficult to determine sequential relationships, and it is plausible that individuals with a lower BMI tend to have poorer sleep quality or that there is a common underlying difference that affects both sleep quality and BMI at the same time. On the other hand, there is also potential for reverse causation as obesity leads to many co-morbidities, including sleep apnea, that can disrupt sleep. While economic hardship may plausibly attenuate the expected link between poor sleep and higher BMI, further research is necessary to clarify these relationships and to avoid overgeneralization based on observational data. The use of self-administered questionnaires may have introduced some degree of error in reporting to parameters like sleep quality, dietary pattern, and psychological behavior. Although the PSQI is a reliable and validated instrument, it cannot be used as an accurate diagnostic tool. While excluding participants with known use of sleep medication strengthens internal validity, it may limit generalizability to populations where sleep medication use is common. Future studies could stratify analyses by medication status or employ objective measures (e.g., actigraphy) to disentangle medication effects.

We noted that the 13% nonresponse may introduce bias, potentially underrepresenting participants with more severe sleep issues or abnormal BMI values. Given the complexity of SEM and the relatively low prevalence of some key variables (e.g., severe depression or stimulant use), this limitation may affect generalizability and the precision of path estimates. While statistical power sufficed to detect mediation effects, the true strength of pathways like sleep → depression → BMI could be misestimated. Caution is warranted in extrapolating results, and future studies should prioritize strategies to improve participation rates for more robust conclusions. Furthermore, the study participants were populations of varying age groups, sexes, educational levels, with different lifestyle habits, and dietary patterns, which may have contributed to inconsistent findings.

Conclusion

This study explored the complex interplay between poor sleep quality, depression, dietary behavior, and BMI among academic staff in Ethiopia, using Structural Equation Modeling (SEM) to disentangle direct and mediated pathways. Contrary to findings from high-income contexts, poor sleep quality demonstrated an inverse association with BMI, mediated primarily through depression and obesogenic dietary behaviors. The unexpected inverse association observed in this study is likely driven by contextual factors unique to low-income settings, such as limited caloric intake, food insecurity, and economic disparity. Figure 2 now visually depicts a potential U-shaped relationship between sleep quality and BMI, supporting the interpretation that both low and high BMI values occur among individuals with poor sleep quality, depending on access to food resources. Results in this study should not be interpreted as a protective metabolic mechanism but rather as a reflection of limited energy intake among sleep-deprived individuals in low-resource settings.

The SEM analysis revealed depression as a critical mediator, bridging sleep quality and dietary choices, while physical activity did not significantly contribute to BMI variation. This challenges conventional energy-balance models and highlights the need for integrated interventions targeting sleep health and mental well-being to mitigate obesity risks, particularly in low-resource settings where food insecurity and stress may dominate metabolic outcomes.

These findings call for integrated interventions that address sleep quality and mental health while considering the broader socioeconomic context. Future studies should employ longitudinal designs and incorporate measures of food security, clinical mental health, and stimulant use to further elucidate the complex interplay between sleep, diet, and weight.

Authors'contributions

Essa A: designed the study, collected and analyzed the data, drafted the manuscript and critically reviewed the manuscript. Getu G: designed the study, collected the data, analyzed the data and reviewed the manuscript. All the authors have read and approved the final manuscript. Alemayehu A: designed the study, supervised the data collection, analyzed the data and reviewed the manuscript.

Funding

The research was conducted with financial funding from the researcher.

Data availability

The datasets used and analyzed during the current study are available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

Ethical approval was obtained from the Institutional Review Board of Jimma University after submission of the proposal. Informed consent to participate was obtained from all of the participants in the study after they were informed about the objective and purpose of the study, including their right not to participate or withdraw at any time. Privacy and confidentiality were maintained both during and after conducting the interviews. For this purpose, all the questionnaires were collected anonymously excluding the name of the respondent.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.