Abstract
Background
Poor sleep quality is prevalent among medical employees and adversely affects their performance and well-being. Limited evidence exists regarding behavioral, psychological, and health-related risk factors for poor sleep quality in medical university staff. We aimed to assess the sleep quality subscales among Iranian medical university employees, and to analyze their association with behavioral, psychological, and health-related factors.
Methods
This cross-sectional analysis utilized baseline data from 1,250 employees at Qazvin University of Medical Sciences who participated in a cohort study conducted between 2021 and 2022. Demographic, behavioral, psychological, and health-related characteristics were assessed in relation to Pittsburgh Sleep Quality Index (PSQI) subscales and effort-reward imbalance (ERI). Participants were selected using a convenience sampling method. Statistical analyses included correlation, chi-square tests, and multinomial Bayesian modeling.
Results
Among participants (56.56% male, mean age 40.57 ± 8.01 years) moderate-to-severe difficulties in sleep duration were most prevalent (54.08%), while sleep medication use was least reported (7.5%). ERI credibly associated with increased odds of sleep difficulties across all PSQI subscales (minimum OR for severe duration sleep quality difficulty = 1.78, 95% CrI: [1.02, 2.90]), except sleep medication use. Depression was associated with all PSQI subscales except sleep latency (minimum OR for severe sleep efficiency difficulty = 2.65, 95% CrI: [1.48, 4.38]). Comorbidity (> 2) was linked to sleep difficulties in specific subscales, including subjective sleep quality, sleep latency, sleep disturbances, and sleep medication use (minimum OR for severe sleep efficiency difficulty = 4.39, 95% CrI: [1.14, 12.28]).
Conclusion
This first population-based study of Iranian medical employees demonstrates that demographic, behavioral, psychological, and health-related factors differentially affect sleep quality subscales, with ERI emerging as a key risk factor. These findings emphasize the importance of tailored interventions to address workplace stress and promote mental and physical health, aiming to enhance sleep quality and overall well-being among medical employees.
Keywords: Pittsburgh sleep quality index (PSQI), Sleep quality subscales, Effort-reward imbalance, Bayesian multinomial logistic regression, Iran
Introduction
Sleep disorders represent a major global public health challenge. These disorders include poor sleep quality, trouble falling or staying asleep, and excessive daytime sleepiness [1, 2]. Poor sleep quality can exacerbate many medical and mental disorders and impair workplace functioning [3]. Furthermore, poor sleep quality is a contributing factor to increased morbidity and mortality and can negatively impact quality of life [4].
Sleep quality is defined as an individual’s self-satisfaction with all aspects of the sleep experience. it includes four key attributes: sleep efficiency, sleep latency, sleep duration, and wake after sleep onset [5]. In addition, the National Sleep Foundation expert panel considers the following indicators of good sleep quality: sleep efficiency, sleep latency, rapid eye movement (REM) and non-REM sleep ratios, naps, arousals, and awakenings [6]. While objective tools like polysomnography and actigraphy provide precise measurements [7, 8], subjective questionnaires are widely used in epidemiological research. The Pittsburgh Sleep Quality Index (PSQI) is one such validated instrument, strongly correlated with objective polysomnographic findings [9].
The prevalence of poor sleep quality is high, particularly among high-stress occupational groups. Epidemiological studies estimate that between 9% and 45% of the general population experiences poor sleep [10, 11], with higher prevalence rates reported among healthcare professionals—58.9% in Ethiopia [12], 39.2% in China [1], and 86.3% in Saudi Arabia [13]. However, sleep disturbance is a critical issue not only for clinical staff but for the entire academic workforce within a medical university. These employees often experience chronic stress from factors such as high workload, administrative pressures, and irregular shift work schedules, all of which may negatively impact their sleep quality [14–16]. Research has demonstrated increased sleep latency and other negative effects on the sleep quality of medical residents [13]. Furthermore, factors such as work-related stress, work-family conflict, and depressive symptoms are consistently identified as key predictors of poor sleep in this population [12, 17, 18]. While social support has been shown to have a protective effect [19, 20], other modifiable risk factors like physical inactivity and smoking are associated with worse sleep outcomes [12, 21].
Although several studies in Iran have assessed overall sleep quality in healthcare settings [19, 22–24], critical gaps remain. First, few have undertaken a comprehensive analysis of the predictors influencing the specific subscales of sleep quality among the broad population of medical university employees. Second, the role of theoretical psychosocial models like Effort-Reward Imbalance [25, 26] in predicting sleep quality subscales in this specific context remains underexplored. This is particularly relevant in Iran, where medical university employees may experience a significant mismatch between their high efforts and received rewards, potentially worsening sleep outcomes. Thus, this study aims to fill these gaps by examining the prevalence of poor sleep quality and its subscales among medical university employees in Iran, and identifying factors associated with each specific sleep quality subscale. The findings will contribute to the literature by providing a nuanced understanding of sleep disturbances in this population and may inform the development of targeted interventions to improve sleep quality and overall well-being.
Methods
Setting and participants
This cross-sectional study utilized baseline data from the Qazvin Employee Health Cohort Study, conducted between 2021 and 2022 to investigate behavioral, psychological, and health-related factors among employees at Qazvin University of Medical Sciences (QUMS) [27].
In Iran, medical universities are integrated health systems responsible for both training healthcare professionals and providing public health services; thus, employees of public hospitals and healthcare centers are university staff. This study’s target population was all permanent employees from the clinical, administrative, educational, and health departments of QUMS with a minimum of two years of employment.
Participants were recruited using a proportional quota sampling method. First, the total number of employees in each university department was determined. Recruitment quotas for each department were then set proportional to its size within the total workforce. Within each department, these quotas were filled by recruiting participants on a voluntary, convenience basis. The sole exclusion criterion was unwillingness to participate. The overall response rate among QUMS employees was 61%. Data from all 1,250 participants who completed the baseline assessment were included in this analysis.
Measurements
Data collection was performed at the cohort center for all enrolled participants through face-to-face interviews by trained research personnel. Informed consent was obtained from all participants prior to their enrollment in the QUMS study.
Demographic variables
Collected data included age, sex, marital status, and educational level. Body mass index (BMI) was calculated from measured weight and height and categorized as normal, overweight, or obese. Wealth status consisted of ownership of assets (e.g., car, freezer, washing machine, microwave, computer) and housing conditions (including number of rooms), as well as number of foreign trips taken. Wealth status was assessed using financial assets and related indicators through principal component analysis [28] (PCA). PCA with varimax rotation was performed to extract three principal components representing the wealth index. The scores from the first principal component were used to rank households, and participants were subsequently categorized into low, moderate, and high wealth tertiles based on these scores.
Behavioral variables
Smoking status was categorized as non-smokers or smokers. Average daily mobile device use was self-reported and categorized into three groups: <3 h, 3–5 h, and ≥ 5 h. Physical activity was assessed using the long form of the International Physical Activity Questionnaire (IPAQ), which comprises 27 items measuring the frequency (days/week) and duration (minutes/day) of vigorous, moderate, and walking activities across four domains: job-related, transportation, household, and leisure-time. Total physical activity volume was calculated as metabolic equivalent of task (MET) minutes per week. Participants were categorized into three physical activity levels based on their total MET-minutes/week as follows: low (< 600 MET-minutes/week), moderate (≥ 600 to < 3000 MET-minutes/week), and high (≥ 3000 MET-minutes/week) [29].
Psychosocial factors
We assessed history of depression through self-reported physician diagnosis. Job stress was assessed using the Effort-Reward Imbalance (ERI) questionnaire developed by Siegrist et al.. (2004). This instrument consists of 22 items rated on a four-point Likert scale ranging from 1 (“completely agree”) to 4 (“completely disagree”), comprising three subscales: effort (5 items), reward (6 items), and over-commitment (11 items). Subscale scores were averaged, and effort-reward ratio (ER ratio) was calculated as effort/(reward*c) where c is a correction factor accounting for the differing number of items [25]. An ER ratio >1 indicates effort-reward imbalance (high job stress). The Persian version of the ERI questionnaire has demonstrated acceptable validity and reliability among employee populations. Internal consistency, assessed using Cronbach’s alpha coefficients, was 0.61 for effort, 0.85 for reward, and 0.67 for over-commitment [30].
Health-related factor
Comorbidities (e.g., ischemic heart disease, hypertension, diabetes, chronic lung disease, thyroid disorders, fatty liver disease, cancers) were obtained by self-report and the total number for each participant was categorized as: zero, one, two, or more than two.
Sleep quality
Sleep quality was assessed using the Pittsburgh Sleep Quality Index (PSQI) [31]. This validated 19-item instrument generates seven component scores: subjective sleep quality, sleep latency, sleep duration, sleep efficiency, sleep disturbances, sleep medications, and daytime dysfunction. Each component is scored on a 0–3 scale, corresponding to no difficulty (0), mild (1), moderate (2), or severe (3) difficulty experienced over a one-month recall period [31]. The Persian version of the PSQI has demonstrated acceptable reliability (Cronbach’s alpha = 0.77) and validity in previous studies [32].
Statistical analysis
Demographic, behavioral, psychological, and health-related factors were summarized using descriptive statistics. Categorical variables were expressed as frequencies and percentages, while continuous variables (e.g., age) were reported as mean ± standard deviation (SD). Sleep difficulty distributions were analyzed across seven PSQI subscale components. Inter-subscale correlations were evaluated using Spearman’s rank correlation.
Given the multinomial nature of these outcomes and the presence of sparse data in some response categories, a Bayesian multinomial regression model was employed for each subscale. This approach was chosen for its ability to provide stable parameter estimates, its intuitive probabilistic interpretation via credible intervals, and its robustness in complex models. Weakly informative priors were used to ensure stable parameter estimation without imposing strong assumptions. Specifically, the intercept terms were assigned normal priors with a mean of 0 and a variance of 4, expressed as α ∼ N (0, 4). Similarly, the regression coefficients were assigned normal priors with the same parameters, βk ∼ N (0, 4). Markov Chain Monte Carlo (MCMC) sampling was conducted with 10,000 draws, including a burn-in period of 4,000 iterations. Results are reported as adjusted odds ratios (ORs) with 95% Bayesian credible intervals (CrIs) (Tables 3, 4, 5, 6, 7, 8 and 9). All subscales are illustrated collectively in Fig. 2 using effect displays. The reference levels for all categorical independent variables are explicitly defined in the tables.
Table 3.
The effect of demographic, behavioral, psychological and Health-Related factors on subjective sleep quality using Bayesian multinomial logistic regressions
| Factors | Difficulty of Subjective sleep quality | |||
|---|---|---|---|---|
| Mild | Moderate | Severe | ||
| OR (95% CrI) | OR (95% CrI) | OR (95% CrI) | ||
| Age | 0.99 (0.97, 1.01) | 0.99 (0.96, 1.02) | 0.98 (0.94, 1.02) | |
| BMI | Underweight | 0.99 (0.28, 2.52) | 2.05 (0.36, 6.38) | 1.92 (0.21, 6.83) |
| Overweight | 0.94 (0.69, 1.27) | 0.94 (0.55, 1.56) | 0.75 (0.32, 1.53) | |
| Obese | 0.92 (0.61, 1.34) | 0.99 (0.51, 1.71) | 0.76 (0.25, 1.79) | |
| Sex | Male | 0.76 (0.56, 1.00) | 0.87 (0.53, 1.37) | 0.37 (0.17, 0.69) |
| Wealth Status | Middle | 1.06 (0.75, 1.44) | 1.04 (0.59, 1.70) | 2.17 (0.99, 4.30) |
| Rich | 0.85 (0.59, 1.15) | 0.81 (0.45, 1.34) | 1.12 (0.45, 2.31) | |
| Marital status | Married | 0.98 (0.63, 1.41) | 0.81 (0.44, 1.39) | 0.75 (0.33, 1.54) |
| Education | Diploma | 0.93 (0.50, 1.61) | 0.75 (0.30, 1.65) | 0.44 (0.13, 1.03) |
| Undergraduate | 0.95 (0.51, 1.61) | 0.69 (0.27, 1.46) | 0.34 (0.10, 0.83) | |
| Postgraduate | 1.38 (0.68, 2.45) | 0.96 (0.32, 2.23) | 0.31 (0.06, 0.91) | |
| Mobile use (hours/day) | 3–5 | 1.06 (0.78, 1.41) | 1.16 (0.69, 1.84) | 1.05 (0.49, 2.01) |
| ≥ 5 | 0.72 (0.44, 1.10) | 1.57 (0.75, 2.93) | 0.75 (0.22, 1.82) | |
| Physical activity | Moderate | 1.18 (0.83, 1.63) | 1.69 (0.95, 2.81) | 0.92 (0.40, 1.92) |
| High | 0.88 (0.60, 1.25) | 0.92 (0.46, 1.63) | 0.99 (0.42, 2.11) | |
| Smoking status | Smoker | 0.94 (0.60, 1.41) | 1.08 (0.51, 1.92) | 1.49 (0.50, 3.49) |
| Effort-Reward | Imbalance | 1.77 (1.33, 2.31) | 2.05 (1.29, 3.18) | 3.40 (1.67, 6.47) |
| Depression | 1.38 (0.76, 2.32) | 2.16 (0.95, 4.22) | 6.62 (2.69, 13.55) | |
| Comorbidity | One | 1.37 (0.97, 1.88) | 1.70 (0.99, 2.67) | 1.67 (0.75, 3.06) |
| Two | 1.65 (0.87, 2.90) | 1.21 (0.34, 2.74) | 7.02 (2.34, 16.73) | |
| > 2 | 2.06 (0.53, 5.80) | 6.54 (1.33, 19.83) | 3.57 (0.34, 13.41) | |
Abbreviation: CrI Bayesian Credible Interval
The reference levels were set as normal (BMI), female (sex), poor (wealth status), single (marital status), under diploma (education), less than 3 h/day (mobile use), low (Physical activity), non-smoker (smoking status), balanced (effort-reward), and zero (comorbidity)
Odds ratios (OR) that are credibly significant are highlighted in bold
Table 4.
The effect of demographic, behavioral, psychological and health-related factors on sleep latency using Bayesian multinomial logistic regressions
| Factors | Difficulty of sleep latency | |||
|---|---|---|---|---|
| Mild | Moderate | Severe | ||
| OR (95% CrI) | OR (95% CrI) | OR (95% CrI) | ||
| Age | 0.99 (0.98, 1.01) | 1.02 (0.99, 1.04) | 1.02 (0.99, 1.05) | |
| BMI | Underweight | 2.44 (0.60, 7.09) | 2.00 (0.17, 8.23) | 2.53 (0.36, 9.50) |
| Overweight | 0.95 (0.69, 1.27) | 0.79 (0.47, 1.28) | 0.50 (0.28, 0.84) | |
| Obese | 0.90 (0.59, 1.31) | 0.42 (0.17, 0.82) | 0.71 (0.34, 1.28) | |
| sex | Male | 0.79 (0.58, 1.05) | 0.63 (0.35, 1.03) | 0.54 (0.29, 0.89) |
| Wealth Status | Middle | 1.13 (0.80, 1.57) | 1.46 (0.80, 2.52) | 1.00 (0.55, 1.67) |
| Rich | 0.87 (0.61, 1.22) | 1.11 (0.58, 2.03) | 0.79 (0.43, 1.34) | |
| Marital status | Married | 0.82 (0.53, 1.22) | 0.40 (0.21, 0.70) | 0.41 (0.21, 0.72) |
| Education | Diploma | 1.04 (0.52, 1.82) | 2.56 (0.76, 6.83) | 0.49 (0.19, 1.03) |
| Undergraduate | 0.91 (0.47, 1.55) | 1.66 (0.50, 4.29) | 0.34 (0.13, 0.69) | |
| Postgraduate | 0.73 (0.37, 1.32) | 1.62 (0.43, 4.39) | 0.10 (0.03, 0.27) | |
| Mobile use (hours/day) | 3–5 | 1.06 (0.78, 1.44) | 1.24 (0.73, 2.00) | 1.50 (0.83, 2.47) |
| ≥ 5 | 0.94 (0.58, 1.47) | 0.65 (0.25, 1.44) | 2.22 (0.97, 4.31) | |
| Physical activity | Moderate | 1.04 (0.73, 1.46) | 1.00 (0.53, 1.74) | 1.16 (0.61, 2.04) |
| High | 0.75 (0.50, 1.07) | 0.95 (0.49, 1.72) | 0.73 (0.36, 1.36) | |
| Smoking status | Smoker | 1.20 (0.76, 1.83) | 2.03 (0.96, 4.00) | 1.47 (0.62, 2.81) |
| Effort-Reward | Imbalance | 1.81 (1.36, 2.37) | 2.56 (1.53, 4.05) | 2.96 (1.75, 4.79) |
| Depression | 0.97 (0.51, 1.67) | 1.90 (0.76, 3.81) | 1.83 (0.78, 3.60) | |
| Comorbidity | One | 1.54 (1.08, 2.13) | 1.00 (0.50, 1.71) | 2.07 (1.15, 3.41) |
| Two | 1.06 (0.56, 1.84) | 1.52 (0.44, 3.44) | 1.79 (0.64, 3.87) | |
| > 2 | 1.87 (0.43, 5.39) | 1.20 (0.08, 4.96) | 6.35 (1.17, 18.98) | |
Abbreviation: CrI Bayesian Credible Interval
The reference levels were set as normal (BMI), female (sex), poor (wealth status), single (marital status), under diploma (education), less than 3 h/day (mobile use), low (Physical activity), non-smoker (smoking status), balanced (effort-reward), and zero (comorbidity)
Odds ratios (OR) that are credibly significant are highlighted in bold
Table 5.
The effect of demographic, behavioral, psychological and Health-Related factors on sleep duration using Bayesian multinomial logistic regressions
| Factors | Difficulty of sleep duration | |||
|---|---|---|---|---|
| Mild | Moderate | Severe | ||
| OR (95% CrI) | OR (95% CrI) | OR (95% CrI) | ||
| Age | 1.00 (0.97, 1.03) | 1.03 (0.99, 1.06) | 1.04 (1.00, 1.07) | |
| BMI | Underweight | 2.69 (0.37, 10.34) | 1.46 (0.18, 5.46) | 1.82 (0.21, 7.05) |
| Overweight | 0.78 (0.41, 1.26) | 0.61 (0.32, 1.01) | 0.67 (0.35, 1.12) | |
| Obese | 1.24 (0.57, 2.60) | 0.73 (0.33, 1.54) | 0.92 (0.41, 1.97) | |
| sex | Male | 1.72 (0.93, 2.81) | 1.45 (0.78, 2.39) | 1.38 (0.73, 2.28) |
| Wealth Status | Middle | 0.86 (0.46, 1.45) | 0.87 (0.47, 1.52) | 0.72 (0.38, 1.22) |
| Rich | 1.10 (0.56, 2.09) | 1.15 (0.56, 2.16) | 0.79 (0.39, 1.49) | |
| Marital status | Married | 1.16 (0.49, 2.23) | 0.95 (0.39, 1.78) | 0.95 (0.39, 1.80) |
| Education | Diploma | 1.71 (0.67, 3.51) | 1.53 (0.59, 3.27) | 1.06 (0.42, 2.17) |
| Undergraduate | 2.59 (1.06, 5.39) | 2.03 (0.79, 4.34) | 1.24 (0.49, 2.53) | |
| Postgraduate | 3.35 (1.05, 8.13) | 2.70 (0.81, 6.82) | 1.79 (0.57, 4.30) | |
| Mobile use (hours/day) | 3–5 | 1.83 (1.01, 3.19) | 1.87 (0.99, 3.30) | 2.18 (1.15, 3.89) |
| ≥ 5 | 0.95 (0.38, 2.09) | 1.25 (0.48, 2.89) | 1.98 (0.76, 4.48) | |
| Physical activity | Moderate | 1.82 (0.98, 3.08) | 1.80 (0.95, 3.11) | 2.09 (1.10, 3.72) |
| High | 1.85 (0.95, 3.26) | 1.41 (0.71, 2.60) | 1.58 (0.78, 2.91) | |
| Smoking status | Smoker | 1.49 (0.60, 3.22) | 1.27 (0.50, 2.71) | 1.12 (0.44, 2.42) |
| Effort-Reward | Imbalance | 0.98 (0.57, 1.57) | 1.39 (0.80, 2.23) | 1.78 (1.02, 2.90) |
| Depression | 0.41 (0.16, 0.89) | 0.92 (0.37, 2.03) | 1.22 (0.51, 2.64) | |
| Comorbidity | One | 1.59 (0.79, 3.16) | 1.84 (0.91, 3.58) | 1.92 (0.92, 3.83) |
| Two | 0.90 (0.29, 2.32) | 1.09 (0.35, 2.86) | 1.48 (0.48, 3.87) | |
| > 2 | 2.19 (0.28, 9.11) | 1.08 (0.09, 4.77) | 4.19 (0.54, 17.98) | |
Abbreviation: CrI Bayesian Credible Interval
The reference levels were set as normal (BMI), female (sex), poor (wealth status), single (marital status), under diploma (education), less than 3 h/day (mobile use), low (Physical activity), non-smoker (smoking status), balanced (effort-reward), and zero (comorbidity)
Odds ratios (OR) that are credibly significant are highlighted in bold
Table 6.
The effect of demographic, behavioral, psychological and Health-Related factors on sleep efficiency using Bayesian multinomial logistic regressions
| Factors | Difficulty of sleep efficiency | |||
|---|---|---|---|---|
| Mild | Moderate | Severe | ||
| OR (95% CrI) | OR (95% CrI) | OR (95% CrI) | ||
| Age | 1.00 (0.97, 1.02) | 1.01 (0.99, 1.04) | 1.02 (1.00, 1.05) | |
| BMI | Underweight | 0.22 (0.01, 1.03) | 1.34 (0.27, 3.87) | 1.34 (0.33, 3.53) |
| Overweight | 1.07 (0.69, 1.60) | 0.78 (0.48, 1.19) | 0.78 (0.53, 1.11) | |
| Obese | 1.06 (0.60, 1.68) | 0.62 (0.32, 1.08) | 0.70 (0.42, 1.10) | |
| sex | Male | 0.96 (0.63, 1.40) | 0.89 (0.56, 1.33) | 0.66 (0.46, 0.92) |
| Wealth Status | Middle | 1.31 (0.81, 2.04) | 1.23 (0.73, 1.91) | 0.91 (0.59, 1.36) |
| Rich | 1.52 (0.91, 2.44) | 0.81 (0.45, 1.32) | 1.12 (0.73, 1.68) | |
| Marital status | Married | 0.75 (0.43, 1.20) | 0.69 (0.39, 1.15) | 0.88 (0.52, 1.40) |
| Education | Diploma | 1.13 (0.44, 2.52) | 0.86 (0.32, 1.92) | 0.55 (0.27, 1.00) |
| Undergraduate | 1.09 (0.43, 2.41) | 0.91 (0.36, 1.99) | 0.46 (0.23, 0.82) | |
| Postgraduate | 1.41 (0.51, 3.26) | 0.85 (0.29, 2.01) | 0.47 (0.21, 0.89) | |
| Mobile use (hours/day) | 3–5 | 0.92 (0.60, 1.33) | 1.43 (0.89, 2.16) | 1.43 (0.97, 2.08) |
| ≥ 5 | 0.80 (0.39, 1.40) | 1.11 (0.49, 2.08) | 1.62 (0.90, 2.74) | |
| Physical activity | Moderate | 1.40 (0.85, 2.23) | 1.24 (0.75, 1.94) | 1.11 (0.73, 1.61) |
| High | 1.02 (0.58, 1.68) | 0.67 (0.36, 1.13) | 0.89 (0.55, 1.36) | |
| Smoking status | Smoker | 0.58 (0.29, 1.04) | 0.55 (0.23, 1.02) | 0.76 (0.41, 1.26) |
| Effort-Reward | Imbalance | 2.28 (1.53, 3.25) | 1.46 (0.95, 2.16) | 1.87 (1.32, 2.60) |
| Depression | 1.62 (0.74, 3.02) | 1.14 (0.42, 2.37) | 2.65 (1.48, 4.38) | |
| Comorbidity | One | 1.14 (0.71, 1.71) | 0.84 (0.46, 1.34) | 1.25 (0.83, 1.80) |
| Two | 1.48 (0.62, 2.94) | 1.02 (0.36, 2.20) | 1.50 (0.70, 2.78) | |
| > 2 | 1.81 (0.21, 6.11) | 1.18 (0.07, 4.62) | 4.39 (1.14, 12.28) | |
The reference levels were set as normal (BMI), female (sex), poor (wealth status), single (marital status), under diploma (education), less than 3 h/day (mobile use), low (Physical activity), non-smoker (smoking status), balanced (effort-reward), and zero (comorbidity)
Abbreviation: CrI Bayesian Credible Interval
Odds ratios (OR) that are credibly significant are highlighted in bold
Table 7.
The effect of demographic, behavioral, psychological and health-Related factors on sleep disturbance using Bayesian multinomial logistic regressions
| Factors | Difficulty of sleep disturbance | |||
|---|---|---|---|---|
| Mild | Moderate | Severe | ||
| OR (95% CrI) | OR (95% CrI) | OR (95% CrI) | ||
| Age | 1.02 (0.99, 1.04) | 1.05 (1.01, 1.09) | 0.81 (0.68, 0.93) | |
| BMI | Underweight | 1.16 (0.25, 3.95) | 5.64 (0.70, 19.77) | 3.34 (0.01, 22.71) |
| Overweight | 1.10 (0.67, 1.69) | 0.81 (0.36, 1.53) | 3.64 (0.13, 14.88) | |
| Obese | 1.60 (0.81, 2.98) | 1.68 (0.59, 3.71) | 3.41 (0.04, 25.95) | |
| Sex | Male | 0.76 (0.48, 1.15) | 0.29 (0.13, 0.52) | 3.63 (0.12, 20.50) |
| Wealth Status | Middle | 0.75 (0.44, 1.18) | 0.69 (0.31, 1.30) | 10.07 (0.12, 49.66) |
| Rich | 1.00 (0.57, 1.63) | 0.82 (0.34, 1.70) | 0.56 (0.01, 2.92) | |
| Marital status | Married | 0.80 (0.40, 1.39) | 0.95 (0.35, 2.12) | 4.54 (0.07, 32.14) |
| Education | Diploma | 1.06 (0.42, 2.17) | 0.40 (0.11, 0.98) | 14.71 (0.17, 98.05) |
| Undergraduate | 1.07 (0.42, 2.12) | 0.34 (0.10, 0.80) | 1.70 (0.04, 10.33) | |
| Postgraduate | 1.18 (0.42, 2.59) | 0.57 (0.14, 1.59) | 1.17 (0.01, 7.77) | |
| Mobile use (hours/day) | 3–5 | 1.42 (0.87, 2.20) | 1.98 (0.87, 3.78) | 0.46 (0.02, 2.45) |
| ≥ 5 | 1.76 (0.80, 3.55) | 4.31 (1.36, 10.57) | 2.53 (0.02, 17.93) | |
| Physical activity | Moderate | 1.13 (0.65, 1.79) | 2.36 (1.00, 4.57) | 2.20 (0.02, 12.98) |
| High | 1.01 (0.56, 1.70) | 2.32 (0.91, 4.92) | 4.57 (0.14, 31.29) | |
| Smoking status | Smoker | 1.17 (0.60, 2.07) | 1.24 (0.38, 2.94) | 37.07 (0.31, 237.99) |
| Effort-Reward | Imbalance | 1.80 (1.16, 2.66) | 3.44 (1.74, 6.14) | 0.49 (0.01, 2.74) |
| Depression | 2.05 (0.71, 5.35) | 5.59 (1.60, 15.17) | 26.52 (0.32, 161.59) | |
| Comorbidity | One | 1.58 (0.90, 2.68) | 2.02 (0.90, 4.14) | 2.51 (0.02, 19.51) |
| Two | 3.97 (0.97, 12.88) | 8.59 (1.47, 32.16) | 4.02 (0.02, 27.40) | |
| > 2 | 0.83 (0.12, 3.32) | 12.31 (1.56, 50.31) | 4.12 (0.02, 29.60) | |
Abbreviation: CrI Bayesian Credible Interval
The reference levels were set as normal (BMI), female (sex), poor (wealth status), single (marital status), under diploma (education), less than 3 h/day (mobile use), low (Physical activity), non-smoker (smoking status), balanced (effort-reward), and zero (comorbidity)
Odds ratios (OR) that are credibly significant are highlighted in bold
Table 8.
The effect of demographic, behavioral, psychological and health-related factors on sleep medication using Bayesian multinomial logistic regressions
| Factors | Difficulty of sleep medication | |||
|---|---|---|---|---|
| Mild | Moderate | Severe | ||
| OR (95% CrI) | OR (95% CrI) | OR (95% CrI) | ||
| Age | 1.02 (0.98, 1.06) | 0.94 (0.89, 1.00) | 1.05 (1.00, 1.10) | |
| BMI | Underweight | 0.46 (0.01, 2.38) | 1.05 (0.01, 6.18) | 2.01 (0.13, 7.78) |
| Overweight | 0.43 (0.19, 0.84) | 0.84 (0.21, 2.18) | 0.71 (0.28, 1.56) | |
| Obese | 0.66 (0.23, 1.41) | 1.23 (0.30, 3.54) | 0.55 (0.14, 1.43) | |
| Gender | Male | 0.61 (0.27, 1.22) | 0.80 (0.21, 2.19) | 1.03 (0.33, 2.60) |
| Wealth Status | Middle | 1.34 (0.54, 3.17) | 1.30 (0.34, 3.48) | 0.64 (0.24, 1.38) |
| Rich | 1.06 (0.39, 2.32) | 1.57 (0.45, 4.42) | 0.80 (0.27, 1.86) | |
| Marital status | Married | 0.64 (0.24, 1.42) | 0.97 (0.25, 2.36) | 0.93 (0.27, 2.64) |
| Education | Diploma | 1.30 (0.32, 3.67) | 2.71 (0.27, 11.91) | 1.06 (0.28, 2.93) |
| Undergraduate | 1.08 (0.29, 2.92) | 1.42 (0.15, 5.94) | 1.07 (0.30, 2.95) | |
| Postgraduate | 0.64 (0.12, 1.98) | 1.58 (0.12, 7.06) | 0.28 (0.02, 1.18) | |
| Mobile use (hours/day) | 3–5 | 0.80 (0.33, 1.64) | 1.16 (0.33, 2.67) | 0.58 (0.21, 1.24) |
| ≥ 5 | 0.86 (0.19, 2.34) | 1.38 (0.27, 4.52) | 0.10 (0.01, 0.42) | |
| Physical activity | Moderate | 1.77 (0.74, 3.69) | 0.90 (0.29, 2.15) | 1.42 (0.44, 3.72) |
| High | 0.65 (0.19, 1.66) | 0.30 (0.05, 0.92) | 0.84 (0.23, 2.23) | |
| Smoking status | Smoker | 2.17 (0.70, 4.96) | 1.19 (0.16, 4.00) | 2.20 (0.62, 5.70) |
| Effort-Reward | Imbalance | 1.45 (0.68, 2.85) | 1.11 (0.36, 2.56) | 1.04 (0.45, 2.05) |
| Depression | 1.25 (0.25, 3.24) | 7.79 (2.09, 19.15) | 12.59 (4.61, 27.88) | |
| Comorbidity | One | 1.63 (0.65, 3.22) | 1.53 (0.34, 3.82) | 0.70 (0.20, 1.63) |
| Two | 4.14 (1.17, 10.60) | 3.00 (0.49, 9.76) | 4.27 (1.25, 10.59) | |
| > 2 | 2.70 (0.13, 11.19) | 0.84 (0.01, 4.67) | 8.00 (1.41, 25.43) | |
Abbreviation: CrI Bayesian Credible Interval
The reference levels were set as normal (BMI), female (sex), poor (wealth status), single (marital status), under diploma (education), less than 3 h/day (mobile use), low (Physical activity), non-smoker (smoking status), balanced (effort-reward), and zero (comorbidity)
Odds ratios (OR) that are credibly significant are highlighted in bold
Table 9.
The effect of demographic, behavioral, psychological and health-related factors on daytime disfunction using Bayesian multinomial logistic regressions
| Factors | Difficulty of daytime disfunction | |||
|---|---|---|---|---|
| Mild | Moderate | Severe | ||
| OR (95% CrI) | OR (95% CrI) | OR (95% CrI) | ||
| Age | 0.99 (0.97, 1.01) | 0.98 (0.96, 1.01) | 0.91 (0.85, 0.96) | |
| BMI | Underweight | 1.73 (0.52, 4.36) | 1.55 (0.18, 5.03) | 0.91 (0.01, 5.09) |
| Overweight | 0.93 (0.68, 1.24) | 1.36 (0.85, 2.07) | 0.57 (0.19, 1.30) | |
| Obese | 0.87 (0.57, 1.24) | 1.65 (0.91, 2.75) | 0.41 (0.06, 1.18) | |
| Sex | Male | 0.95 (0.69, 1.25) | 0.69 (0.45, 1.02) | 0.73 (0.24, 1.69) |
| Wealth Status | Middle | 1.00 (0.71, 1.38) | 0.80 (0.49, 1.23) | 3.28 (0.94, 8.56) |
| Rich | 1.00 (0.69, 1.40) | 0.65 (0.39, 1.03) | 1.08 (0.25, 2.93) | |
| Marital status | Married | 1.00 (0.66, 1.44) | 1.17 (0.65, 2.02) | 3.60 (0.88, 10.71) |
| Education | Diploma | 1.02 (0.56, 1.72) | 1.07 (0.42, 2.31) | 0.74 (0.06, 3.27) |
| Undergraduate | 1.10 (0.62, 1.87) | 1.49 (0.60, 3.18) | 1.03 (0.17, 3.95) | |
| Postgraduate | 1.37 (0.72, 2.40) | 1.42 (0.52, 3.05) | 1.83 (0.16, 7.89) | |
| Mobile use (hours/day) | 3–5 | 1.15 (0.85, 1.52) | 1.16 (0.75, 1.66) | 2.91 (1.01, 7.66) |
| ≥ 5 | 1.36 (0.82, 2.09) | 1.50 (0.73, 2.62) | 3.98 (0.79, 11.73) | |
| Physical activity | Moderate | 0.97 (0.68, 1.34) | 1.12 (0.66, 1.73) | 0.83 (0.25, 2.18) |
| High | 0.79 (0.53, 1.13) | 0.81 (0.45, 1.30) | 0.93 (0.20, 2.58) | |
| Smoking status | Smoker | 1.36 (0.87, 2.03) | 1.22 (0.60, 2.18) | 1.05 (0.11, 3.29) |
| Effort-Reward | Imbalance | 1.61 (1.22, 2.09) | 2.06 (1.33, 3.03) | 2.29 (0.82, 5.51) |
| Depression | 2.58 (1.39, 4.54) | 4.00 (1.84, 7.76) | 14.95 (4.25, 38.69) | |
| Comorbidity | One | 1.06 (0.75, 1.45) | 1.05 (0.65, 1.63) | 1.68 (0.53, 3.86) |
| Two | 0.91 (0.50, 1.54) | 0.57 (0.19, 1.22) | 1.43 (0.08, 5.41) | |
| > 2 | 0.58 (0.15, 1.47) | 0.65 (0.09, 2.10) | 3.09 (0.11, 14.17) | |
Abbreviation: CrI Bayesian Credible Interval
The reference levels were set as normal (BMI), female (sex), poor (wealth status), single (marital status), under diploma (education), less than 3 h/day (mobile use), low (Physical activity), non-smoker (smoking status), balanced (effort-reward), and zero (comorbidity)
Odds ratios (OR) that are credibly significant are highlighted in bold
Fig. 2.
Bayesian multinomial logistic regression results: association of demographic, behavioral, psychological, and health-related factors with all seven PSQI sleep subscales
To assess the convergence of the Markov Chain Monte Carlo algorithm for the seven sleep subscales, we used trace plots and computed the Gelman-Rubin statistic. A value below 1.1 was considered indicative of convergence. All diagnostic plots and statistics indicated that the algorithm had reached its target distribution for all parameters. All analyses were conducted using R software version 4.4.3.
Results
This cross-sectional analysis included 1,250 employees from QUMS. Table 1 presents the demographic, behavioral, psychosocial, and health-related characteristics of the participants. The sample comprised 707 (56.56%) males and 543 (43.44%) females, with a mean age of 40.57 ± 8.01 years. The study revealed that 46.36% of participants were overweight, 49.39% reported daily mobile phone use exceeding three hours, 14.72% were unmarried, and 71.28% held an academic degree. Participants were evenly distributed across wealth tertiles. More than two-thirds of the participants (78.08%) had moderate or high physical activity. Regarding occupational factors, 60.61% reported effort-reward imbalance at work. Additionally, 7.02% reported a history of depression, and nearly one-third reported at least one comorbid condition.
Table 1.
Demographic, behavioral, psychosocial, and health-related factors
| Variables | Total (n = 1250) | |
|---|---|---|
| Demographic Factors | ||
| Age | 40.57 ± 8.01 | |
| BMI | Underweight | 20 (1.69%) |
| Normal | 386 (32.66%) | |
| Overweight | 548 (46.36%) | |
| Obese | 228 (19.29%) | |
| Sex | Female | 543 (43.44%) |
| Male | 707 (56.56%) | |
| Wealth status | Poor | 412 (33.36%) |
| Middle | 411 (33.28%) | |
| Rich | 412 (33.36%) | |
| Marital status | Single | 184 (14.72%) |
| Married | 1066 (85.28%) | |
| Education | Under diploma | 87 (6.96%) |
| Diploma | 272 (21.76%) | |
| Undergraduate | 685 (54.8%) | |
| Postgraduate | 206 (16.48%) | |
| Behavioral Factors | ||
| Mobile use (hours/day) | < 3 | 625 (50.61%) |
| 3–5 | 473 (38.3%) | |
| ≥ 5 | 137 (11.09%) | |
| Physical activity | Low | 271 (21.93%) |
| Moderate | 572 (46.28%) | |
| High | 393 (31.8%) | |
| Smoking status | Non-smokers | 1054 (88.05%) |
| Smokers | 143 (11.95%) | |
| Psychosocial Factors | ||
| Effort-Reward Imbalance | Balance | 488 (39.39%) |
| Imbalance | 751 (60.61%) | |
| Depression | No | 1139 (92.98%) |
| Yes | 86 (7.02%) | |
| Health-Related Factor | ||
| Comorbidity | Zero | 877 (70.16%) |
| One | 277 (22.16%) | |
| Two | 78 (6.24%) | |
| > two | 18 (1.44%) | |
The results are reported as n (%) for qualitative variable and mean ± SD for quantitative
Preliminary analyses included correlation and chi-square tests to examine initial relationships between variables. Table 2 presents associations between PSQI sleep components and ERI, depression, and comorbidity based on chi-square tests. Significant associations were found between ERI and all sleep subscales (P-value < 0.05) except sleep medication use (P-value = 0.881). Depression showed significant associations with all seven sleep components (P-value < 0.05). Comorbidity was significantly associated with four subscales: subjective sleep quality (P-value = 0.004), sleep latency (P-value = 0.005), sleep disturbances (P-value = 0.008), and sleep medication use (P-value = 0.001).
Table 2.
Distribution of psychological and health-related factors across PSQI subscales and results of chi-squared tests for independence
| χ | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Subscales | Level of Difficulty | Effort-Reward Imbalance | Comorbidity Count | Depression | ||||||||
| Balance (n = 488) | Imbalance (n = 751) | P-value | Zero (n = 877) | One (n = 277) | Two (n = 78) | > two (n = 18) | P-value | No (n = 1139) | Yes (n = 86) | P-value | ||
| Subjective sleep quality | No | 225 (46.2%) | 246 (32.98%) | < 0.001 | 352 (40.6%) | 93 (33.7%) | 25 (32.05%) | 4 (22.22%) | 0.004 | 448 (39.51%) | 21 (24.71%) | < 0.001 |
| Mild | 208 (42.71%) | 369 (49.46%) | 400 (46.14%) | 135 (48.91%) | 38 (48.72%) | 7 (38.89%) | 536 (47.27%) | 36 (42.35%) | ||||
| Moderate | 41 (8.42%) | 87 (11.66%) | 83 (9.57%) | 34 (12.32%) | 6 (7.69%) | 5 (27.78%) | 109 (9.61%) | 13 (15.29%) | ||||
| Severe | 13 (2.67%) | 44 (5.9%) | 32 (3.69%) | 14 (5.07%) | 9 (11.54%) | 2 (11.11%) | 41 (3.62%) | 15 (17.65%) | ||||
| Sleep Latency | No | 210 (43.12%) | 209 (28.02%) | < 0.001 | 310 (35.76%) | 81 (29.35%) | 28 (35.9%) | 3 (16.67%) | 0.005 | 391 (34.48%) | 22 (25.88%) | 0.011 |
| Mild | 216 (44.35%) | 376 (50.4%) | 407 (46.94%) | 147 (53.26%) | 33 (42.31%) | 8 (44.44%) | 553 (48.77%) | 37 (43.53%) | ||||
| Moderate | 30 (6.16%) | 77 (10.32%) | 81 (9.34%) | 17 (6.16%) | 8 (10.26%) | 1 (5.56%) | 93 (8.2%) | 11 (12.94%) | ||||
| Severe | 31 (6.37%) | 84 (11.26%) | 69 (7.96%) | 31 (11.23%) | 9 (11.54%) | 6 (33.33%) | 97 (8.55%) | 15 (17.65%) | ||||
| Sleep Duration | No | 39 (8.01%) | 47 (6.3%) | 0.009 | 65 (7.5%) | 13 (4.71%) | 7 (8.97%) | 1 (5.56%) | 0.115 | 77 (6.79%) | 8 (9.41%) | < 0.001 |
| Mild | 211 (43.33%) | 269 (36.06%) | 348 (40.14%) | 105 (38.04%) | 24 (30.77%) | 6 (33.33%) | 462 (40.74%) | 16 (18.82%) | ||||
| Moderate | 130 (26.69%) | 206 (27.61%) | 237 (27.34%) | 79 (28.62%) | 20 (25.64%) | 2 (11.11%) | 307 (27.07%) | 26 (30.59%) | ||||
| Severe | 107 (21.97%) | 224 (30.03%) | 217 (25.03%) | 79 (28.62%) | 27 (34.62%) | 9 (50%) | 288 (25.4%) | 35 (41.18%) | ||||
| Sleep Efficiency | No | 304 (62.42%) | 380 (51.01%) | 0.001 | 492 (56.81%) | 151 (54.71%) | 40 (51.28%) | 6 (33.33%) | 0.053 | 647 (57.11%) | 32 (37.65%) | < 0.001 |
| Mild | 53 (10.88%) | 122 (16.38%) | 122 (14.09%) | 39 (14.13%) | 12 (15.38%) | 2 (11.11%) | 159 (14.03%) | 14 (16.47%) | ||||
| Moderate | 50 (10.27%) | 85 (11.41%) | 102 (11.78%) | 27 (9.78%) | 6 (7.69%) | 1 (5.56%) | 127 (11.21%) | 8 (9.41%) | ||||
| Severe | 80 (16.43%) | 158 (21.21%) | 150 (17.32%) | 59 (21.38%) | 20 (25.64%) | 9 (50%) | 200 (17.65%) | 31 (36.47%) | ||||
| Sleep Disturbance | No | 65 (13.35%) | 64 (8.58%) | 0.003 | 104 (12%) | 22 (7.97%) | 4 (5.13%) | 1 (5.56%) | 0.008 | 127 (11.2%) | 3 (3.53%) | < 0.001 |
| Mild | 395 (81.11%) | 616 (82.57%) | 712 (82.12%) | 231 (83.7%) | 64 (82.05%) | 7 (38.89%) | 934 (82.36%) | 64 (75.29%) | ||||
| Moderate | 26 (5.34%) | 66 (8.85%) | 50 (5.77%) | 23 (8.33%) | 10 (12.82%) | 10 (55.56%) | 73 (6.44%) | 17 (20%) | ||||
| Severe | 1 (0.21%) | 0 (0%) | 1 (0.12%) | 0 (0%) | 0 (0%) | 0 (0%) | 0 (0%) | 1 (1.18%) | ||||
| Use of Sleep Medication | No | 455 (93.43%) | 690 (92.49%) | 0.881 | 818 (94.35%) | 255 (92.39%) | 65 (83.33%) | 13 (72.22%) | 0.001 | 1070 (94.36%) | 63 (74.12%) | < 0.001 |
| Mild | 13 (2.67%) | 26 (3.49%) | 23 (2.65%) | 10 (3.62%) | 5 (6.41%) | 1 (5.56%) | 34 (3%) | 3 (3.53%) | ||||
| Moderate | 7 (1.44%) | 11 (1.47%) | 10 (1.15%) | 6 (2.17%) | 2 (2.56%) | 0 (0%) | 12 (1.06%) | 6 (7.06%) | ||||
| Severe | 12 (2.46%) | 19 (2.55%) | 16 (1.85%) | 5 (1.81%) | 6 (7.69%) | 4 (22.22%) | 18 (1.59%) | 13 (15.29%) | ||||
| Daytime Dysfunction | No | 232 (47.64%) | 259 (34.72%) | < 0.001 | 338 (38.99%) | 108 (39.13%) | 38 (48.72%) | 10 (55.56%) | 0.416 | 470 (41.45%) | 16 (18.82%) | < 0.001 |
| Mild | 193 (39.63%) | 359 (48.12%) | 397 (45.79%) | 121 (43.84%) | 31 (39.74%) | 5 (27.78%) | 501 (44.18%) | 43 (50.59%) | ||||
| Moderate | 55 (11.29%) | 112 (15.01%) | 118 (13.61%) | 40 (14.49%) | 8 (10.26%) | 2 (11.11%) | 147 (12.96%) | 19 (22.35%) | ||||
| Severe | 7 (1.44%) | 16 (2.14%) | 14 (1.61%) | 7 (2.54%) | 1 (1.28%) | 1 (5.56%) | 16 (1.41%) | 7 (8.24%) | ||||
The results are reported by n (%)
Associations between all qualitative variables were evaluated using the chi-squared (χ²) test, with significant associations highlighted in bold
Figure 1 illustrates the distribution of sleep difficulties across the seven subscales and their correlations. The highest prevalence of moderate-to-severe difficulty observed for sleep duration (54.08%), while the lowest prevalence was found for sleep medication use (7.5%). The strongest correlation was between sleep duration and sleep efficiency (r = 0.73, P-value < 0.001), while the weakest correlations were between daytime dysfunction and both sleep latency (r = 0.13, P-value < 0.001) and sleep efficiency (r = 0.14, P-value < 0.001).
Fig. 1.
Difficulty distribution of sleep subscales and the heatmap of correlations
Bayesian multinomial logistic regression analysis was conducted to examine associations between predictor variables and sleep quality subscales. Figure 2 provides an overview of these effects, with detailed results presented in Tables 3, 4, 5, 6, 7, 8 and 9. Odds ratios (ORs) in Tables 3, 4, 5, 6, 7, 8 and 9 represent the relative odds of experiencing “mild”, “moderate” or “severe” problems compared to “no problem” for each sleep subscale. Reference categories for independent variables were: normal BMI, female sex, poor wealth status, single marital status, under-diploma education, < 3 h/day mobile use, low physical activity, non-smoker status, balanced ERI, and zero comorbidities.
The results indicate that older age was not associated with difficulties in subjective sleep quality or sleep latency. It was associated with higher odds of moderate sleep disturbances (OR = 1.05, 95% CrI: [1.01, 1.09]) and severe sleep medication use (OR = 1.05, 95% CrI: [1.00, 1.10]), but with lower odds of severe daytime dysfunction (OR = 0.91, 95% CrI: [0.85, 0.96]) and moderate sleep medication use (OR = 0.94, 95% CrI: [0.89, 1.00]). Trends were observed for severe difficulties in sleep duration and efficiency.
Overweight participants, compared to those with a normal BMI, were associated with lower odds of severe sleep latency difficulties (OR = 0.50, 95% CrI: [0.28, 0.84]) and mild sleep medication use (OR = 0.43, 95% CrI: [0.19, 0.84]). Similarly, obese participants were associated with lower odds of moderate sleep latency difficulties (OR = 0.42, 95% CrI: [0.17, 0.82]) (Tables 4 and 8, respectively).
Males showed significantly lower odds than females of severe difficulties in subjective sleep quality (OR = 0.37, 95% CrI: [0.17, 0.69]), sleep latency (OR = 0.54, 95% CrI: [0.29, 0.89]), and sleep efficiency (OR = 0.66, 95% CrI: [0.46, 0.92]), as well as moderate sleep disturbances (OR = 0.29, 95% CrI: [0.13, 0.52]) (Tables 3, 4, 6 and 7, respectively).
The analysis revealed that, compared to low physical activity, high physical activity was associated with higher odds of severe sleep duration difficulties (OR = 2.09, 95% CrI: [1.10, 3.72]) and moderate physical activity was associated with 70% lower odds of moderate sleep medication use (OR = 0.30, 95% CrI: [0.05, 0.92]).
Marital and smoking status were generally not associated with sleep difficulties. The exception was marital status, where married participants had lower odds of both moderate (OR = 0.40, 95% CrI: [0.21, 0.70]) and severe (OR = 0.41, 95% CrI: [0.21, 0.72]) sleep latency problems compared to single participants (Table 4).
Effort-reward imbalance was credibly associated with difficulties in all sleep components, except for sleep medication use (Table 8). Participants in the imbalanced group consistently showed higher odds of sleep difficulties in all sleep subscales compared to the balanced group.
Higher education levels were generally protective across most sleep subscales, except for mild sleep duration difficulties where higher education was associated with increased odds (OR = 3.35, 95% CrI: [1.05, 8.13]) (Table 5). Furthermore, higher education was not associated with sleep medication use or daytime dysfunction.
A history of depression was associated with higher odds of difficulties in all sleep components except for sleep latency. The strongest associations were observed for severe subjective sleep quality (OR = 6.62, 95% CrI: [2.69, 13.55]), severe sleep medication use (OR = 12.59, 95% CrI: [4.61, 27.88]), and severe daytime dysfunction (OR = 14.95, 95% CrI: [4.25, 38.69]).
Mobile device use was associated with several sleep difficulties. Using a mobile for 3–5 h daily was associated with higher odds of severe sleep duration difficulties (OR = 2.18, 95% CrI: [1.15–3.89]) and severe daytime dysfunction (OR = 2.91, 95% CrI: [1.01–7.66]). Using a mobile for ≥ 5 h daily was associated with higher odds of moderate sleep disturbances (OR = 4.31, 95% CrI: [1.36–10.57]) but with lower odds of severe sleep medication use (OR = 0.10, 95% CrI: [0.01–0.42]).
A higher comorbidity burden was associated with increased odds of severe sleep difficulties. Participants with more than two comorbidities had substantially higher odds of severe sleep latency difficulties (OR = 6.35, 95% CrI: [1.17, 18.98]), severe sleep efficiency difficulties (OR = 4.39, 95% CrI: [1.14, 12.28]), and severe sleep medication use (OR = 8.00, 95% CrI: [1.41, 25.43]) compared to those with no comorbidities. No credible associations were found between comorbidity status and either sleep duration (Table 5) or daytime dysfunction (Table 9).
Discussion
This cross-sectional study provides a comprehensive evaluation of the associations between a wide range of factors and the seven subscales of all seven sleep quality subscales among medical university employees in Iran, using Bayesian multinomial logistic regressions.
Sleep disturbances represent a significant public health concern, potentially exacerbating existing health conditions and impairing occupational functioning [1]. The PSQI offers a robust framework for multidimensional sleep assessment, capturing specific subscales such as sleep latency, duration, efficiency, disturbances, use of sleep medication, and daytime dysfunction [22]. By focusing on these subscales, our findings provide novel insights into the predictors of sleep quality subscales within this occupational group.
Our multivariate analysis revealed differential associations between various factors and the PSQI subscales.
Subjective sleep quality
Analysis revealed that being male, and higher education levels were associated with lower odds of reporting severe difficulties. Conversely, ERI and depression were strongly associated with higher odds of impairment.
While the associations for depression and comorbidity align with established literature [33], the significant association with ERI provides a more nuanced understanding of psychosocial work stress. This finding is similar to the results of Güngördü et al. [34], who also identified significant work stress using a different construct (workload-control-support). Our results specifically suggest that an imbalance between effort and reward is a key predictor of impaired subjective sleep quality, highlighting the importance of addressing reward equity, such as through recognition programs, in workplace policies.
Sleep latency
Being female was associated with higher odds of severe difficulties, while being married and having higher education levels were associated with lower odds. Furthermore, compared to those with a normal BMI, overweight participants had lower odds of severe sleep latency difficulties. ERI was a consistent predictor, associated with higher odds of sleep latency difficulties across all severity levels. A higher comorbidity burden was also associated with increased odds of severe difficulties; notably, participants with more than two conditions had substantially higher odds compared to those with none.
The finding that a greater number of chronic diseases is associated with prolonged sleep latency aligns with Hsu et al. who proposed that higher number of diseases are linked to increase the difficulty in falling asleep or prolonged sleep latency [35].
Sleep duration
Older age and male sex were associated with higher odds of severe difficulties. Moderate physical activity and ERI were also associated with higher odds of this outcome. Furthermore, mobile device usage of 3–5 h per day was associated with increased odds of severe sleep duration difficulties compared to lower use.
The association between digital device use and sleep duration difficulties is supported by existing literature on technology-mediated sleep disruption [36]. The finding that higher physical activity levels were associated with sleep duration difficulties adds to a complex body of evidence; while some studies report benefits [37], others note null [38] or even negative associations [39]. These inconsistencies could be due to factors such as exercise timing [40] and age-related factors [41].
We found no significant association for comorbidity status, a result consistent with the prior findings of Hsu et al. [35].
Sleep efficiency
Older age, being female, ERI, depression, and higher comorbidity count were associated with increased odds of severe difficulties, while higher education was associated with lower odds.
The association between being female and poorer sleep efficiency aligns with the findings of Raju et al. [42], and the strong link with depression is consistent with established literature on its debilitating effects on poor sleep efficiency [43].
A novel finding of our study is the significant association between ERI and impaired sleep efficiency. This contrasts with the null results reported by Jackowska et al. [44] in their female-only cohort, suggesting that ERI’s effect on sleep may be more pronounced in a gender-mixed occupational population. Furthermore, our finding that comorbidity burden impairs sleep efficiency contrasts with the null findings of with Hsu et al.‘s [35], potentially indicating population-specific differences.
Sleep disturbance
Older age, being female, mobile device use exceeding five hours per day, ERI, depression, and a greater comorbidity burden were all associated with higher odds of sleep disturbances.
Our results reinforce and extend existing literature in two key areas. First, the association between ERI and sleep disturbances aligns with the work of Fahlen et al. [45]. Second, we found that high mobile device use was associated with disturbances. While prior research has often focused on youth [46, 47], our study confirms that this association extends to medical employees across adult age groups. Furthermore, we found a significant association between comorbidity burden and sleep disturbances, which contrasts with the null findings reported by Hsu et al. [48].
Sleep medication use
Older age, depression, and a greater comorbidity burden were all associated with increased odds of severe sleep medication use. Conversely, and unexpectedly, high mobile device use (≥ 5 h daily) was associated with lower odds of severe sleep medication use.
This finding regarding comorbidity burden differs from the null results reported by Hsu et al. [35]. Collectively, these findings underscore the complex interplay between psychological factors (depression), physical health (comorbidity), and behavioral patterns (mobile use) in understanding sleep medication use.
Daytime dysfunction
Older age was associated with lower odds of daytime dysfunction. In contrast, mobile device use of ≥ 5 h per day, ERI, and a history of depression were associated with increased odds of impairment.
The association with depression is consistent with findings from Liu et al. [49]. A novel contribution of our study is the identification of ERI as a significant predictor of daytime dysfunction, highlighting it as a previously under-investigated occupational risk factor for daytime impairment. This suggests that workplace interventions targeting psychosocial stress, such as efforts to reduce effort-reward imbalance, may be beneficial for improving daytime alertness.
In summary, this study highlights important factors that affect each of the seven PSQI sleep quality subscales in medical employees. These findings offer a basis for creating tailored sleep interventions in medical settings and guide future research on sleep health at work.
Strengths and limitations
This study benefits from a large sample, comprehensive multidimensional sleep assessment, and Bayesian modeling that stabilizes estimates despite sparse data in severe categories. However, several limitations must be acknowledged. The cross-sectional design precludes causal inferences, and self-reported measures may introduce bias. we encountered limited sample sizes across some sleep subscale categories, resulting in data sparsity issues particularly for severe difficulty classifications. While we addressed this through Bayesian analytical methods with weakly informative priors to stabilize estimates, these statistical adjustments cannot fully compensate for small subgroup numbers and we have wide credible interval in some of them. The convenience quota sampling limits generalizability beyond medical university employees. Residual confounding, particularly from unmeasured occupational roles or circadian factors, remains possible. Future studies with longitudinal design, objective sleep measures, and stratified occupational analyses are needed. We recommend future studies employ: (1) substantially larger samples to ensure adequate representation across all subscale severity levels, (2) complementary objective sleep measures to validate self-reports, (3) longitudinal designs to establish temporal relationships and causal pathways and (4) consider important confounders like occupational status.
Conclusion
This study provides the first comprehensive analysis of the multifaceted determinants of sleep quality among Iranian medical employees, revealing that demographic, behavioral, psychological, and health-related factors distinctly influence each PSQI subscale. Notably, effort-reward imbalance consistently emerged as a prominent risk factor across nearly all sleep subscales, underscoring the pervasive impact of workplace stress on sleep health in this occupational group. These findings advance current knowledge by identifying both established and novel predictors-such as the unique role of ERI-thereby offering actionable insights for targeted interventions and workplace policies aimed at improving sleep quality among healthcare professionals. While Bayesian methods mitigated data sparsity limitations, the study’s reliance on self-reported measures and modest sample sizes across subscales suggests the need for future research with larger, more diverse cohorts incorporating objective sleep measurements. Despite these limitations, our findings show how workplace stress and mental health affect sleep in medical staff. These results can help hospitals and medical schools inform the development of targeted programs to improve sleep, which could benefit both workers and patient care. Furthermore, our findings provide valuable evidence based for developing targeted workplace interventions to improve sleep health among medical employees. The identified associations between occupational stressors, psychological factors, and sleep disturbances offer concrete pathways to enhance both staff well-being and healthcare system performance.
Acknowledgements
The authors wish to thank the Vice-Chancellor of Research and Technology of Qazvin University of Medical Sciences for the financial support of the study. Also, we thank all employee participants for their helpful cooperation.
Abbreviations
- REM
Rapid eye movement
- PSQI
Pittsburgh Sleep Quality Index
- QUMS
Qazvin University of Medical Sciences
- BMI
Body mass index
- ERI
Effort-Reward Imbalance
- SD
Standard deviation
- OR
Odds ratios
Authors' contributions
ZH and NM developed and designed the research project overall. NM led the analysis and the writing of the manuscript and ZH, EK and MH contributed to the analysis and co-wrote the manuscript. All authors read and approved the final manuscript.
Funding
The Qazvin University of Medical Sciences funded the study. The funders did not have a role in study design, data collection, analysis, reporting, or the decision to submit for publication.
Data availability
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.
Declarations
Ethics approval and consent to participate
The Research Ethics Committee of Qazvin University of Medical Sciences approved this study (Ethics Code: IR.QUMS.REC.1404.057). Written informed consent was obtained from all participants prior to study implementation. Anonymous data were kept confidential and used according to the Declaration of Helsinki.
Competing interests
The authors declare no competing interests.
Consent for publication
Not applicable.
Footnotes
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
- 1.Qiu D, Yu Y, Li R-Q, Li Y-L, Xiao S-Y. Prevalence of sleep disturbances in Chinese healthcare professionals: a systematic review and meta-analysis. Sleep Med. 2020;67:258–66. [DOI] [PubMed] [Google Scholar]
- 2.Kalmbach DA, Arnedt JT, Song PX, Guille C, Sen S. Sleep disturbance and short sleep as risk factors for depression and perceived medical errors in first-year residents. Sleep. 2017;40(3):zsw073. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Deng X, Liu X, Fang R. Evaluation of the correlation between job stress and sleep quality in community nurses. Medicine. 2020;99(4):e18822. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Robbins R, Quan SF, Weaver MD, Bormes G, Barger LK, Czeisler CA. Examining sleep deficiency and disturbance and their risk for incident dementia and all-cause mortality in older adults across 5 years in the United States. Aging. 2021;13(3):3254. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Nelson KL, Davis JE, Corbett CF, editors. Sleep quality: an evolutionary concept analysis. Nursing forum. Wiley Online Library; 2022. [DOI] [PubMed]
- 6.Ohayon M, Wickwire EM, Hirshkowitz M, Albert SM, Avidan A, Daly FJ, et al. National sleep foundation’s sleep quality recommendations: first report. Sleep Health. 2017;3(1):6–19. [DOI] [PubMed] [Google Scholar]
- 7.Åkerstedt T, Hume K, Minors D, Waterhouse J. The meaning of good sleep: a longitudinal study of polysomnography and subjective sleep quality. J Sleep Res. 1994;3(3):152–8. [DOI] [PubMed] [Google Scholar]
- 8.Jean-Louis G, Kripke DF, Mason WJ, Elliott JA, Youngstedt SD. Sleep estimation from wrist movement quantified by different actigraphic modalities. J Neurosci Methods. 2001;105(2):185–91. [DOI] [PubMed] [Google Scholar]
- 9.Buysse DJ, Reynolds CF, Monk TH, Hoch CC, Yeager AL, Kupfer DJ. Quantification of subjective sleep quality in healthy elderly men and women using the Pittsburgh sleep quality index (PSQI). Sleep. 1991;14(4):331–8. [PubMed] [Google Scholar]
- 10.Chen P, Lam MI, Si TL, Zhang L, Balbuena L, Su Z, et al. The prevalence of poor sleep quality in the general population in China: a meta-analysis of epidemiological studies. Eur Arch Psychiatry Clin Neurosci. 2024;274(7):1–14. [DOI] [PubMed] [Google Scholar]
- 11.Madrid-Valero JJ, Martínez-Selva JM, Couto BRd, Sánchez-Romera JF, Ordoñana JR. Age and gender effects on the prevalence of poor sleep quality in the adult population. Gac Sanit. 2017;31:18–22. [DOI] [PubMed] [Google Scholar]
- 12.Tesfaye W, Getu AA, Dagnew B, Lemma A, Yeshaw Y. Poor sleep quality and associated factors among healthcare professionals at the university of Gondar comprehensive specialized hospital, Northwest Ethiopia. Front Psychiatry. 2024;15:1225518. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.AlSaif HI. Prevalence of and risk factors for poor sleep quality among residents in training in KSA. J Taibah Univ Med Sci. 2019;14(1):52–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Ferri P, Guadi M, Marcheselli L, Balduzzi S, Magnani D, Di Lorenzo R. The impact of shift work on the psychological and physical health of nurses in a general hospital: a comparison between rotating night shifts and day shifts. Risk Manag Healthc Policy. 2016;203:11. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Kumar A, Bhat PS, Ryali S. Study of quality of life among health workers and psychosocial factors influencing it. Indian J Psychiatry. 2018;27(1):96–102. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Janatolmakan M, Naghipour A, Khatony A. Prevalence and factors associated with poor sleep quality among nurses in COVID-19 wards. Sci Rep. 2024;14(1):16616. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Luo Y, Fei S, Gong B, Sun T, Meng R. Understanding the mediating role of anxiety and depression on the relationship between perceived stress and sleep quality among health care workers in the COVID-19 response. Nat Sci Sleep. 2021. 10.2147/NSS.S313258. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Mohamed BES, Ghaith RFAH, Ahmed HAA. Relationship between work–family conflict, sleep quality, and depressive symptoms among mental health nurses. Middle East Curr Psychiatry. 2022;29(1):19. [Google Scholar]
- 19.Haseli A, Egdampur F, Qaderi K, Kaffashian MR, Delpisheh A. Nurses and midwives’ sleep quality and its associated factors during the early COVID-19 pandemic in Iran. Heliyon. 2023. 10.1016/j.heliyon.2023.e15068. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Xiao H, Zhang Y, Kong D, Li S, Yang N. The effects of social support on sleep quality of medical staff treating patients with coronavirus disease 2019 (COVID-19) in January and February 2020 in China. Med Sci Monitor: Int Med J Experimental Clin Res. 2020;26:e923549–1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Alghamdi LA, Alsubhi LS, Alghamdi RM, Aljahdaly NM, Barashid MM, Wazira LA, et al. Prevalence of poor sleep quality among physicians and nurses in a tertiary health care center. J Taibah Univ Med Sci. 2024;19(3):473–81. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Nikfar B, Moazzami B, Chaichian S, Ghalichi L, Ekhlasi-Hundrieser M, Chashmyazdan M, et al. Sleep quality and its main determinants among staff in a Persian private hospital. Arch Iran Med. 2018;21(11):524–9. [PubMed] [Google Scholar]
- 23.Mansouri P, Rivaz M, Soltani R, Tabei P. The relationship between professional quality of life and sleep quality among nurses working in intensive care units: a cross-sectional study. BMC Nurs. 2025;24(1):34. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Behbahani HB, Borazjani F, Sheikhi L, Amiri R, Angali KA, Nejad SB et al. The association between diet quality scores with sleep quality among employees: a cross-sectional study. Ethiop J Health Sci. 2022;32(1). [DOI] [PMC free article] [PubMed]
- 25.Siegrist J, Starke D, Chandola T, Godin I, Marmot M, Niedhammer I, et al. The measurement of effort–reward imbalance at work: European comparisons. Soc Sci Med. 2004;58(8):1483–99. [DOI] [PubMed] [Google Scholar]
- 26.Siegrist J. Adverse health effects of high-effort/low-reward conditions. J Occup Health Psychol. 1996;1(1):27. [DOI] [PubMed] [Google Scholar]
- 27.Health cohort of employees of Qazvin University of Medical Sciences [cited 2025]. Available from: https://irancohorts.ir/qazvin-cohort-study/
- 28.Vyas S, Kumaranayake L. Constructing socio-economic status indices: how to use principal components analysis. Health Policy Plann. 2006;21(6):459–68. [DOI] [PubMed] [Google Scholar]
- 29.Ashok P, Kharche JS, Raju R, Godbole G. Metabolic equivalent task assessment for physical activity in medical students. Natl J Physiol Pharm Pharmacol. 2017;7(3):236. [Google Scholar]
- 30.Yadegarfar G, Alinia T, Hosseini R, Hassannejad R, Fayaz M, Sanati J, et al. Psychometric properties of the Farsi version of effort-reward imbalance questionnaire: a longitudinal study in employees of a synthetic fibre factory in Iran. Int Arch Occup Environ Health. 2013;86:147–55. [DOI] [PubMed] [Google Scholar]
- 31.Buysse DJ, Reynolds CF, Monk TH, Berman SR, Kupfer DJ. The pittsburgh sleep quality index: a new instrument for psychiatric practice and research. Psychiatry Res. 1989;28(2):193–213. 10.1016/0165-1781(89)90047-4. [DOI] [PubMed] [Google Scholar]
- 32.Farrahi Moghaddam J, Nakhaee N, Sheibani V, Garrusi B, Amirkafi A. Reliability and validity of the Persian version of the Pittsburgh sleep quality index (PSQI-P). Sleep Breath. 2012;16:79–82. [DOI] [PubMed] [Google Scholar]
- 33.Hsu M-F, Lee K-Y, Lin T-C, Liu W-T, Ho S-C. Subjective sleep quality and association with depression syndrome, chronic diseases and health-related physical fitness in the middle-aged and elderly. BMC Public Health. 2021;21(1):164. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Güngördü N, Kurtul S, Erdoğan MS. Evaluation of sleep quality, work stress and related factors in hospital office workers. Med Lav. 2023;114(3):e2023023. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Hsu M-F, Lee K-Y, Lin T-C, Liu W-T, Ho S-C. Subjective sleep quality and association with depression syndrome, chronic diseases and health-related physical fitness in the middle-aged and elderly. BMC Public Health. 2021;21:1–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Akowuah PK, Nti AN, Ankamah-Lomotey S, Frimpong AA, Fummey J, Boadi P, et al. Digital device use, computer vision syndrome, and sleep quality among an African undergraduate population. Adv Public Health. 2021;2021(1):6611348. [Google Scholar]
- 37.Kredlow MA, Capozzoli MC, Hearon BA, Calkins AW, Otto MW. The effects of physical activity on sleep: a meta-analytic review. J Behav Med. 2015;38:427–49. [DOI] [PubMed] [Google Scholar]
- 38.Atoui S, Chevance G, Romain A-J, Kingsbury C, Lachance J-P, Bernard P. Daily associations between sleep and physical activity: a systematic review and meta-analysis. Sleep Med Rev. 2021;57:101426. [DOI] [PubMed] [Google Scholar]
- 39.Kakinami L, O’Loughlin EK, Brunet J, Dugas EN, Constantin E, Sabiston CM, et al. Associations between physical activity and sedentary behavior with sleep quality and quantity in young adults. Sleep Health. 2017;3(1):56–61. [DOI] [PubMed] [Google Scholar]
- 40.Wendt A, da Silva ICM, Gonçalves H, Menezes A, Barros F, Wehrmeister FC. Short-term effect of physical activity on sleep health: a population-based study using accelerometry. J Sport Health Sci. 2022;11(5):630–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Li L, Yu Q, Zhao W, Herold F, Cheval B, Kong Z, et al. Physical activity and inhibitory control: the mediating role of sleep quality and sleep efficiency. Brain Sci. 2021;11(5):664. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Raju A, Nithiya DR, Tipandjan A. Relationship between burnout, effort-reward imbalance, and insomnia among informational technology professionals. J Educ Health Promot. 2022;11(1):296. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Steiger A, Pawlowski M. Depression and sleep. Int J Mol Sci. 2019;20(3):607. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.Jackowska M, Dockray S, Hendrickx H, Steptoe A. Psychosocial factors and sleep efficiency: discrepancies between subjective and objective evaluations of sleep. Psychosom Med. 2011;73(9):810–6. [DOI] [PubMed] [Google Scholar]
- 45.Fahlén G, Knutsson A, Peter R, Åkerstedt T, Nordin M, Alfredsson L, et al. Effort–reward imbalance, sleep disturbances and fatigue. Int Arch Occup Environ Health. 2006;79:371–8. [DOI] [PubMed] [Google Scholar]
- 46.Thomée S, Härenstam A, Hagberg M. Mobile phone use and stress, sleep disturbances, and symptoms of depression among young adults-a prospective cohort study. BMC Public Health. 2011;11:1–11. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47.Liu S, Wing YK, Hao Y, Li W, Zhang J, Zhang B. The associations of long-time mobile phone use with sleep disturbances and mental distress in technical college students: a prospective cohort study. Sleep. 2019;42(2):zsy213. [DOI] [PubMed] [Google Scholar]
- 48.Marin-Alejandre BA, Abete I, Cantero I, Riezu-Boj JI, Milagro FI, Monreal JI, et al. Association between sleep disturbances and liver status in obese subjects with nonalcoholic fatty liver disease: a comparison with healthy controls. Nutrients. 2019;11(2):322. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 49.Liu Y, Peng T, Zhang S, Tang K. The relationship between depression, daytime napping, daytime dysfunction, and snoring in 0.5 million Chinese populations: exploring the effects of socio-economic status and age. BMC Public Health. 2018;18:1–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.