Introduction

Post-stroke depression (PSD) is the most common neuropsychological disorder and may lead to burdensome consequences among stroke survivors. This adversely affects recovery, cognitive function, and survival among patients with [29]. According to the Diagnostic and Statistical Manual of Mental Disorders V (DSM-V), PSD is a mood disorder caused by general health problems (such as stroke), with specific features including depressive symptoms, severe depression-like episodes, manic symptoms, or mixed characteristics [1]. In 1955, Martin Roth et al. performed the first experimental studies on PSD and revealed the association between stroke and depression [25]. Furthermore, PSD was shown to be a phenomenon that may become a chronic condition. This neuropsychiatric condition may lead to reduced independence in daily activities, prolonging rehabilitation time and even causing suicidal thoughts and mortality [23].

A systematic review of 61 observational studies indicated that the frequency of PSD is 33%, ranging from 29 to 36% [12]. The significant fluctuation of incidence rates is mostly attributed to methodological differences, various depression rating scales, and different diagnostic criteria for patient recruitment. In recent decades, overwhelming evidence has identified a set of related factors of post-stroke depression, such as gender, demography, level of handicap, global medical history, mental disorder history, and stroke characteristics [26, 29]. In particular, post-stroke fatigue was determined as an important impact factor in PSD [24]. Post-stroke fatigue is often relieved when depression is appropriately treated [9].

One phenomenon that has attracted the attention of recent scholars is the association between PSD and sleep changes [22, 27] due to its nature as a modifiable factor. In fact, 20–40% of stroke survivors have reported sleep disorders [22], which significantly influenced post-stroke outcomes and recovery. Furthermore, several sleep disturbances such as the sleep–wake cycle, including long and short sleep duration, circadian rhythm disorders, and insomnia were considered the most strongly associated factors with depression [18]. Thus, numerous depression screening tools accepted sleep problems as a component of the scale. Although previous analyses using the global Pittsburgh Sleep Quality Index (PSQI) score explored the presence of a correlation between PSD and sleep disorders [6, 13], it remains unclear whether or not each PSQI component has an impact on depression among patients with stroke. Consequently, further studies are essential to explore the association between sleep quality components and post-stroke depression status.

Despite the extensive work published about the prevalence of post-stroke depression and related factors around the world, little information is available about this figure in low or middle-income countries such as Vietnam. Furthermore, it is unknown if risk factors such as fatigue or PSQI scale components have effects on PSD. Therefore, this study aimed to determine the prevalence of post-stroke depression (PSD) during the first year and its associated factors, especially focusing on sleep quality and fatigue severity.

Materials and methods

Design, setting, and sampling

A cross-sectional study was conducted among stroke survivors at the National Geriatric Hospital (Hanoi, Vietnam) from July to December 2021. The criteria for inclusion in this study were as follows: (1) aged ≥ 45 and (2) having had a confirmed stroke diagnosis one year prior. People who experienced (1) neurological diseases, (2) preexisting mood disorders, (3) preexisting cognitive dysfunction before the stroke, (4) using antidepressants, anxiolytics, or antipsychotics, (5) having transient ischemic attacks, (6) unstable medical conditions, (7) who were unable to give their informed consent, and (8) cases of consciousness disorders such as coma and poor cognition were excluded. We used a formula for estimating a population proportion with relative precision to calculate the sample size as follows: n = Z21−α/2 * [p*(1 − p)/\(\varepsilon\) 2], with detailed indicators, including the required sample size (n), Z1 − α/2 = 1.96 (with α = 0.05% and 95% confidence interval), the prevalence of depression in post-stroke patients (p), and relative precision assumed as 0.11 (\(\varepsilon\)). A recent study on stroke survivors in Vietnam showed that 69.6% of patients have depression [17], thus we chose p = 0.696 for the formula. The sample size was determined to have at least 139 participants. The sample size was increased by 10% to avoid incomplete responses or dropouts. Therefore, we recruited a total of 157 stroke patients, which provided sufficient data for the analysis. Patients with stroke were directly interviewed and evaluated using standardized questionnaires. Data collection was implemented by five trained nurses to ensure the reliability of assessment results. The nurses were trained on how to use the questionnaires and screen patients eligible to participate in the study. A pilot study was performed with 10 patients with stroke to determine the suitability of sentences and grammar utilized. Data from the pilot sample was not included in the current study.

Measure and instruments

Outcome variables

Post-stroke depressive symptoms

The Patient Health Questionnaire-9 (PHQ-9) belongs to the PRIME-MD diagnostic instrument for common mental illnesses, a self-administered questionnaire. It has been shown by previous studies that the PHQ-9 scale is a valid and reliable instrument to assess depression in the Vietnamese population [20, 33]. This instrument scores each item according to criteria as “0” (not at all) to “3” (nearly every day). It can be used to make a preliminary diagnosis of depression for people who are at high risk for depression, such as those who had a stroke [15]. In the total score of 0–27, the cut-off point used was 10 [15]. Cronbach’s alpha for items related to answers in our study was good at 0.87.

Covariables

Demographic variables

The questions answered by the interviewees include age, gender (male/female), education (secondary school and lower/high school and upper), occupation (not working/working), and marital status (living with spouse/single, widow, or widower).

Stroke-related characteristics

Stroke characteristics were collected based on aspects including (i) stroke classification (ischemia/hemorrhage/unknown), which was based on medical records, (ii) time from stroke onset to study participation (< 1 month/1 month to < 3 months/3 months to < 6 months/6 months to 1 year), (iii) hemispheric lesion locations (left-sided lesions/right-sided lesions/unknown), which were categorized by the affected vascular territory through magnetic resonance imaging (MRI) or computed tomography (CT) scan, (iv) receiving beta-blocker therapy (yes/no), and (v) disability of stroke survivors according to Modified Rankin Scale (MRS) with six levels (no symptoms (0 points)/no significant disability (1 point)/slight disability (2 points)/moderate disability (3 points)/moderate-severe disability (4 points)/severe disability (5 points)).

Post-stroke fatigue

Post-stroke fatigue severity was evaluated by the Fatigue Severity Scale (FSS), which has high validity for stroke patients [10]. The FSS is a short self-reported questionnaire that contains 10 statements about the severity of fatigue symptoms. The scoring method is based on a seven-point Likert scale, with a value of 1 indicating “strongly disagree” and a value of 7 indicating “strongly agree.” The cut-off point for determining whether or not a person is non-fatigued or fatigued after stroke was 4 in the overall score of 0–70 [10]. In our study, Cronbach’s alpha for items regarding answers was good at 0.89.

Instrumental activities of daily living

The respondents were interviewed based on eight domains of function as determined by the Instrumental Activities of Daily Living (IADL). This instrument is most helpful for determining the functional activity levels of stroke patients at present as well as evaluating improvement or deterioration over time. Men are not assessed in the domains of food preparation, housekeeping, or laundry, while women are scored in all eight areas of function. The participants are scored at their highest level of performance in each domain [11]. A summary score ranged from 0 (poor function, dependent) to 8 (excellent function, independent).

Sleep quality

The 19-item Pittsburgh Sleep Quality Index (PSQI) was used to measure the subjective experience of sleep quality in the previous month. This instrument comprises seven components (subjective sleep quality, sleep latency, sleep duration, habitual sleep efficiency, sleep disturbance, use of sleep medication, and daytime dysfunction) that are combined from 19 separate items. Each item is scored on a 0–3 interval scale, giving a total score ranging from 0 to 21 points. A lower score indicates better sleep quality. Previous research has shown that a global PSQI score of over 5 may be used to discriminate between good and bad sleepers with high sensitivity and specificity [3]. A previous study indicated that the Vietnamese version of PSQI is a valid and reliable tool that can be used for assessing sleep disorders in Vietnamese patients or for community screening [30].

Data analysis

STATA version 16 (StataCorp. LP, College Station, Texas, USA) was utilized to address the study’s main objectives. Variables such as sociodemographic and clinical characteristics were analyzed by descriptive statistics. The mean and standard deviation (SD) were used to represent continuous variables, while frequencies and percentages were utilized to describe categorical variables. To determine depression categorization, the PHQ-9 scores of each participant were used. Whitney Wilcoxon test was employed for continuous variables and the χ2 test was used for categorical variables. These aid in detecting a statistically significant difference between the groups with and without depressive symptoms.

Univariate regressions were used before using the multivariable regression model according to the spirit of the “purposeful selection of variables in regression methods” suggested by Hosmer and Lemeshow [14]. This could be to aid in selecting the predictors to include in the basic multivariable regression model. Thus, using univariate logistic regression models, the unadjusted association between PSD (with PHQ-9 score > 10) and predicted variables was determined. Besides, those without depression (PHQ-9 score ≤ 10) served as the reference group [19]. The multivariate logistic regression model comprised all variables that were significant in the unadjusted models, and logistic regression was performed to determine predictors associated with PSD (yes = 1/no = 0). Stepwise forward methods help reduce and filter these models with a p-value > 0.2 as the threshold for excluding variables. Statistical significance was defined as a p-value \(\le\) 0.05.

Results

Table 1 shows the sociodemographic characteristics of the participants according to classification with or without depressive symptoms. The mean age of the 157 stroke patients was 73.1 (SD = ± 9.6) years, with 48% of them being female, 67% of the participants had no career at the time of the stroke, and the participants responded that their education level of high school and higher accounted for 56%. Most of the participants lived within spousal relationships (84%). A statistically significant difference between with and without post-stroke depressive symptoms was found in the variable of education levels (secondary school and lower/high school and upper) with p-value \(\le\) 0.05.

Table 1 Patient sociodemographic characteristics according to classification with or without post-stroke depressive symptoms
Full size table

Table 2 demonstrates the clinical features of the participants according to classification with or without depressive symptoms. The percentage of PSD accounted for 38% of the participants. The respondents reported stroke status due to ischemia (64%), hemorrhage (115%), and unknown cause (22%). The majority of the participants were categorized into slight to moderate-severe disability groups (54%). The prevalence of post-stroke depression was highest in the first 3-month period following stroke onset (60%). The group with defined depressive symptoms had an average time from stroke onset to study participation of about 96 days (3 months). The majority of the participants had lesions in the left hemispheric location (n = 68, 43%). Among those with post-stroke depressive symptoms, left-sided lesions were prominent, with a percentage of 47%. Furthermore, 33 (21%) patients received beta-blocker therapy. Most of the post-stroke patients had fatigue (62%) and dependent activities of daily living (57%). The differences in post-stroke fatigue (including FSS scores), activities of daily living (including IADL scores, and disability of stroke survivors between the two groups of “no depressive symptoms” and “depressive symptoms” were statistically significant (p \(\le\) 0.05).

Table 2 Patient clinical characteristics according to classification with or without post-stroke depressive symptoms
Full size table

Table 3 shows the patient sleep scores and components of the PSQI scale according to depression status. The global score and PSQI components of the participants with PSD showed higher levels than those without depressive symptoms. Except for use of sleep medication, all the remaining PSQI components and global scores were related to depression classification (p-value \(\le\) 0.05).

Table 3 Patient sleep scores and components of the PSQI scale according to classification with or without post-stroke depressive symptoms
Full size table

In the logistic multivariable model adjusted for other associated symptoms (Table 4), the factors associated with the PSD groups were higher FSS scores (OR = 5.51; 95% CI: 1.74; 17.40) and higher scores in two domains of the PSQI scale, including subjective sleep quality (OR = 3.03; 95% CI: 1.21; 7.58) and sleep disturbance (OR = 5.22; 95% CI: 1.33; 20.47) with p-value \(\le\) 0.05.

Table 4 Associated factors with post-stroke depressive symptoms through the logistic regression analysis
Full size table

Discussion

This study assessed the association between PSD and related factors in Vietnam, especially focusing on the patients’ sleep quality during the first year after stroke. The results indicated that the prevalence of PSD accounted for 38% of the respondents. This figure focused on groups that have high functional disability levels, daily living activity impairment, and post-stroke fatigue, with a statistically significant difference (p-value \(\le\) 0.05). In particular, most of the participants who had experienced a stroke for the first 3 months reported a high rate of depressive symptoms (60%). All the PSQI scale components and global scores showed poorer sleep quality among the PSD group. The multivariate logistic regression analysis explored post-stroke fatigue and two significant components of the PSQI scale (including subjective sleep quality and sleep disturbance) associated significantly with depressive symptoms after stroke.

This study discovered a significant prevalence of post-stroke depression, which is consistent with prior studies that revealed a cumulative figure of depression for 1 year after stroke ranging from 36 to 41% [4, 16, 32]. However, compared to a systematic review and meta-analysis in 2014, the percentage of PSD was synthesized from 61 studies, which was lower than the current finding of 33% (ranging from 29 to 36%) in the first year after stroke [12]. In Vietnam, there is only one study that mentioned this figure [17]. We found that the prevalence of depression (based on a PHQ-9 score ≥ 10) [15], accounted for 43% in the first month following stroke onset and 17% in the 1- to 3-month period. Our findings are higher than the pooled prevalence of 61 observational studies according to a meta-analysis in 2014, with 28% in the 0- to 1-month period after an occurring stroke, and 36% in the period from 2–5 months [12]. Furthermore, the previous study found that post-stroke depression ranged from 14 to 63% at 3 months [28]. However, it is complicated to analyze these pooled estimates due to the heterogeneity between studies. Furthermore, our study revealed that the prevalence of PSD was higher in patients with low IADL scores and functional disability at high levels. This result was consistent with previous studies that identified a meaningful association between PSD and functional disability [21, 31]. The association between daily living activity impairment and depression reached the highest level at 6 months and then decreased; however, it significantly remained at 1 year after stroke [21]. The common appearance of PSD in patients with severe disability and low IADL scores might reflect its impact on retarding recovery. Therefore, routine depression screening is the best way to prevent and decrease the negative impact on further recovery and quality of life for stroke survivors.

The significant association between depressive symptoms and post-stroke fatigue in the participants was explored through a logistic regression analysis. In the literature, researchers suggested fatigue as the most common concomitant symptom related to depression in stroke patients [9, 24]. Both symptoms often coexist and show similar experiences, making them difficult to distinguish as independent events [5]. From biomedical aspects, previous researchers explained PSD as a multifactorial psychological phenomenon [2]. Furthermore, the strong connection between post-stroke depression, sleep disturbance, and fatigue could be considered as evidence to support the multifactorial perspective of PSD [7, 27]. As a result, post-stroke fatigue conditions should be attended to by healthcare professionals and caregivers in both acute and chronic phases. Early fatigue screening will positively affect recovery and reduce worse outcomes when they are present.

The present findings revealed that those with post-stroke depressive symptoms had higher PSQI global scores than the remaining group, which is consistent with a previous study by Davis et al. [6]. Sleep disorders have been identified as a significant risk factor for depression in both the acute and chronic stages of stroke [8, 34]. In the acute stage, neuroendocrine changes and biological triggers (e.g., inflammatory biomarkers) may have a more significant influence than sleep quality in patients with stroke [34]. However, sleep quality may continue to affect depression in chronic stroke survivors over time. Changes in a person’s brain activity, as well as the macrostructure and microstructure of their sleep, may contribute to depression in patients with chronic stroke [8]. Previous study indicated that sleep circadian rhythm and an individual’s sleep–wake cycle may be changed by lesions caused by a stroke. These alterations may result in a variety of sleep disorders, which can contribute to depression [8]. In fact, sleep aids physical energy restoration and has good effects on mental functions such as memory consolidation. Stroke survivors can in turn disturb their sleep-rest patterns if levels of depression increase [6]. When the seven PSQI subcomponents were examined independently, we discovered that the “subjective sleep quality” and “sleep disturbance” (difficulty maintaining sleep) domains were associated with PSD among the participants. Previous research also showed that poor subjective sleep quality among stroke survivors was significantly associated with depressive symptoms [6]. Likewise, sleep disturbances delay stroke recuperation and might even be related to the development of depressive symptoms after stroke [8]. Therefore, focusing on these two modifiable sleep characteristics will be a promising intervention strategy for reducing depressive symptoms in those who are suffering from chronic stroke. Simultaneously, early diagnosis and treatment of sleep disturbances and poor sleep quality are essential to ensure the quality of life and reduce the risk of psychiatric problems after stroke.

The strengths of our study are that it was the first work in Vietnam to explore the association of all PSQI scale components and related factors with post-stroke depression. The PSQI scale has been shown to be a reliable tool with high sensitivity and specificity, and may be utilized for evaluating sleep problems in Vietnamese patients with stroke or for community screening [30]. Nevertheless, several study limitations should be considered in light of the current findings. First, sleep disorders and fatigue are considered to be risk factors for depression [5], which may occur before or after stroke. Since a cross-sectional study design was implemented in this survey, it did not allow us to establish cause-and-effect relationships according to temporal interaction between sleep quality/post-stroke fatigue and PSD. Thus, in the future, longitudinal studies should be implemented to examine whether or not screening and treating high-risk post-stroke survivors with depression, fatigue, and sleep disturbance is effective. Second, this is a single hospital-based cross-sectional study with a small sample size, so it may not be possible to extrapolate the results to the entire Vietnamese stroke population. Third, PHQ-9 has been shown to be a depression-measuring instrument with good sensitivity and specificity for stroke survivors. However, diagnosis of PSD should be based on the clinical standards of mental health professionals instead of PHQ-9. This may exclude false-positive cases due to the complex symptoms of stroke. Fourth, data collection was implemented from July to December 2021, which is in the COVID-19 pandemic era. Consequently, it could not be excluded that depressive symptoms occurred in the changing context of the COVID-19 pandemic. Fifth, although the FSS is the most frequently used scale in stroke, validation studies on the Vietnamese version are lacking. However, this scale showed an excellent internal consistency for participant response items in the current investigation, with a Cronbach’s alpha of 0.89. Hence, this scale may be useful for post-stroke fatigue screening.

Conclusion

This investigation displayed the high prevalence of PSD and revealed that PSD was more common in patients with impairment of daily living activity and high levels of functional disability. Thus, routine depression screening may be useful to assess in order to intervene to decrease the negative effect on further recovery and quality of life for stroke survivors. Fatigue and two components of the PSQI scale (subjective sleep quality and sleep disturbance) were found to have an association with post-stroke depression.