Sex Differences in the Prevalence and Correlates of Post-Stroke Insomnia: A Population-Based Study of 7,147 Indonesian Stroke Survivors

2.1 Study sample

The data utilized in this study were obtained from the 2018 Basic Health Research (Riset Kesehatan Dasar, RISKESDAS) conducted by the Indonesian Ministry of Health. The RISKESDAS survey is a thorough national evaluation conducted quinquennially to collect health-related data. The National Institute of Health Research and Development (NIHRD) administers it under the Ministry of Health in Indonesia. The survey was administered by enumerators possessing formal education in health, holding at least health diplomas. The enumerators performed face-to-face interviews utilizing validated questions for accuracy and dependability. Furthermore, they performed measurements and assessments during home visits, as outlined in the supplementary document issued by the Ministry of Health in Bahasa Indonesia (Kementerian Kesehatan, 2019). Members of each household (those residing on the premises for at least six months and sharing the same food income source) were solicited to partake in the survey. Health-related data on patients diagnosed with stroke was obtained from the 2018 RISKESDAS dataset.

2.2 Inclusion and exclusion criteria

We only considered participants who satisfied the requirements listed below in order to meet the inclusion criteria: 1) having received a stroke diagnosis in the past; and 2) having submitted the necessary information for the insomnia variable. Participants whose insomnia variable had missing values were not included in the study.

2.3 Dependent variable

A question related to insomnia was examined: “In the past two weeks, have you experienced any sleep disturbances, such as difficulty initiating sleep, nocturnal awakenings, or early morning arousal?” The insomnia status was assumed if participants answered ‘yes’.

2.4 Independent variables

2.4.1 Sociodemographic variables

Sociodemographic characteristics encompassed age, gender, educational attainment, marriage status, occupation, and degree of urbanization. We classified educational attainment into three categories: ‘no formal education’ (0), ‘non-university level’ (1), and ‘university level’ (2). Marital status was classified as ‘single’ (0) or ‘married’ (1), and ‘widow or widower’ (2). Occupation status was categorized as ‘unemployed’ (0) and ‘employed’ (1), and urbanization level as ‘rural’ (0) and ‘urban’ (1).

2.4.2 Health and lifestyle behavior variables

In this research, we incorporated health conditions and lifestyle behaviors, including regular follow-up, comorbidities (such as cancer, asthma, cardiovascular disease, and chronic renal illness), vigorous and moderate physical activity, smoking status, and alcohol consumption.

The evaluation of regular follow-up for stroke was conducted using the subsequent inquiry: Has your stroke state been consistently monitored by a physician at the healthcare facility? the potential responses of “yes,” ‘sometimes,’ and ‘no.’

Comorbidities were identified based on the response (‘yes’ = 0 or ‘no’ = 1) to the question: ‘Have you received a diagnosis of (….) from a physician?’

Smoking status was assessed by the inquiries: ‘Are you a smoker?’ and ‘Have you smoked in the past month?’ Smoking status was classified as ‘non-smoker’ (code = 0), ‘occasional smoker’ (code = 1), and ‘daily smoker’ (code = 2).

Alcohol intake was evaluated by the inquiry: ‘Have you ingested alcohol in the past month?’ with response options of ‘yes’ and ‘no.’

Physical activity was evaluated utilizing the modified Global Physical Activity Questionnaire (GPAQ) from the World Health Organization for surveillance purposes. The GPAQ gathers data on physical activity in three areas: occupational or domestic activities, transportation, and leisure pursuits. All activities must have been conducted uninterrupted for a minimum of 10 minutes per week. Participants were requested to enumerate activities they typically engaged in daily, categorizing them as ‘yes’ for activities performed and ‘no’ for those not performed, for both vigorous and moderate physical activity. The assessment of GPAQ in Indonesia demonstrated satisfactory inter-rater reliability for categorical variables, with kappa values between 0.44 and 0.78, and test-retest reliability for continuous variables, with Spearman’s rho ranging from 0.45 to 0.80.⁷

2.5 Ethical consideration and informed consent statement

Ethical approval for the original RISKESDAS 2018 data collection was granted by the Ethical Committee of Health Research, NIHRD, Ministry of Health, Republic of Indonesia (No. LB.02.01/2/KE.267/2017). Secondary analysis for the present study was approved by the Health Research Ethics Committee of the Faculty of Medicine, Universitas Halu Oleo (No. 62/UN29.20.2/ETIK/2025). Written informed consent was obtained from all participants in the original survey. As this study uses only de-identified secondary data, no additional consent was required.

2.6 Statistical analysis

Statistical analysis was conducted utilizing SPSS (Version 29, SPSS Inc., Chicago, IL, USA). Descriptive statistics were analyzed for demographic features and aggregated responses, encompassing the mean and standard deviation for continuous variables and percentages for categorical data. Subsequently, we performed analyses utilizing χ2 (Chi-square) for categorical variables and the Mann-Whitney U-test (given the data’s deviation from a normal distribution) for continuous variables to investigate the presence of gender disparities in sociodemographic characteristics, health conditions, and lifestyle behaviors.

Subsequently, forward and stepwise logistic regression analyses were conducted to examine the impact of sociodemographic variables, health problems, and lifestyle behaviors on stroke populations with and without insomnia, accounting for gender differences. Hosmer–Lemeshow goodness-of-fit tests were performed to evaluate the models’ alignment with the observed outcome. The alpha criterion for hypothesis testing was established at p < 0.05.

3.1 Study characteristic

At the beginning, 713,783 individuals were screened for eligibility. Among them, 705,741 participants were excluded because they did not have a stroke diagnosis. Furthermore, 895 participants were excluded because they had missing values on the insomnia variable. The final analysis comprised 7,147 stroke participants ( Figure 1).

Figure 1. Participant flow chart.

n = number of participants.

Table 1 presents a comparison of participants’ characteristics between males and females. The insomnia cohort exhibited a greater mean age than the non-insomnia cohort across both genders. Among males, educational attainment, work position, urbanization, consistent follow-up, and the presence of comorbidities such as cancer, heart disease, asthma, and participation in regular physical exercise exhibited significant differences between the insomnia and non-insomnia groups (all P < 0.05). Among females, educational attainment, marital status, and the occurrence of comorbidities such as cardiovascular disease and asthma exhibited significant differences between the insomnia and non-insomnia cohorts (all P < 0.05). Additional demographic information is available in Table 1 and Table 2.

3.2 Prevalence of post-stroke insomnia

The overall prevalence of PSI among stroke survivors in Indonesia was 28.4%. Figure 2 depicts the comparative prevalence of PSI among males and females across different age groups. Our findings demonstrate that the incidence of insomnia increases with age. We grouped the age into 13 age groups, ranging from 15–19 years old to more than 75 years old. There is a trend of increasing prevalence in the male insomnia group, ranging from 25% to 32%. Similarly, we also find an increasing trend in insomnia prevalence for the female age group (26% to 37%). Overall, we found that females have a higher trend in PSI prevalence compared to males (25.6% vs 31.2%, respectively).

Figure 2. Post-stroke insomnia prevalence by gender and age group.

3.3 Analysis of regression

The univariate regression analyses are displayed in Table 3 and Table 4. Multivariate regression analyses are displayed in Table 5 and Table 6. Active smoking elevates the probability of insomnia development in comparison to non-smokers among subjects. Individuals with a greater educational attainment (university level) are less prone to developing insomnia. Individuals residing in urban environments are less prone to sleeplessness than those inhabiting rural regions. Individuals who partake in physical activity are less prone to developing insomnia. Female participants with comorbidities, including heart disease and asthma, exhibited a higher propensity for developing insomnia.