Development of insomnia in patients with stroke: A systematic review and meta-analysis

Background and aim Stroke is a serious threat to human life and health, and post-stroke insomnia is one of the common complications severely impairing patients’ quality of life and delaying recovery. Early understanding of the relationship between stroke and post-stroke insomnia can provide clinical evidence for preventing and treating post-stroke insomnia. This study was to investigate the prevalence of insomnia in patients with stroke. Methods The Web of Science, PubMed, Embase, and Cochrane Library databases were used to obtain the eligible studies until June 2023. The quality assessment was performed to extract valid data for meta-analysis. The prevalence rates were used a random-efect. I2 statistics were used to assess the heterogeneity of the studies. Results Twenty-six studies met the inclusion criteria for meta-analysis, with 1,193,659 participants, of which 497,124 were patients with stroke. The meta-analysis indicated that 150,181 patients with stroke developed insomnia during follow-up [46.98%, 95% confidence interval (CI): 36.91–57.18] and 1806 patients with ischemic stroke (IS) or transient ischemic attack (TIA) developed insomnia (47.21%, 95% CI: 34.26–60.36). Notably, 41.51% of patients with the prevalence of nonclassified stroke developed insomnia (95% CI: 28.86–54.75). The incidence of insomnia was significantly higher in patients with acute strokes than in patients with nonacute strokes (59.16% vs 44.07%, P < 0.0001). Similarly, the incidence of insomnia was significantly higher in the patients with stroke at a mean age of ≥65 than patients with stroke at a mean age of <65 years (47.18% vs 40.50%, P < 0.05). Fifteen studies reported the follow-up time. The incidence of insomnia was significantly higher in the follow-up for ≥3 years than follow-up for <3 years (58.06% vs 43.83%, P < 0.05). Twenty-one studies used the Insomnia Assessment Diagnostic Tool, and the rate of insomnia in patients with stroke was 49.31% (95% CI: 38.59–60.06). Five studies used self-reporting, that the rate of insomnia in patients with stroke was 37.58% (95% CI: 13.44–65.63). Conclusions Stroke may be a predisposing factor for insomnia. Insomnia is more likely to occur in acute-phase stroke, and the prevalence of insomnia increases with patient age and follow-up time. Further, the rate of insomnia is higher in patients with stroke who use the Insomnia Assessment Diagnostic Tool.


Introduction
Stroke is the second most morbid and deadly disease globally, which is characterized by high morbidity, disability, mortality, and recurrence.It substantially threatens human life, health, and quality of life [1,2].Previous study revealed that neuropsychiatric disorders frequently affect stroke survivors, such as insomnia, depression, or anxiety and so on [3].Similarly, One third of stroke patients met the diagnostic criteria of insomnia, and patients may experience difficulty falling asleep, difficulty with sleep persistence, and early awakening [4].
Insomnia is the most common sleep disorder prevalent in people of all ages.In severe cases, it can affect daytime work and life, and even cause emotional disorders [5].The incidence of insomnia increases with the increase in social pressure [6].Study showed that the incidence of insomnia in stroke patients is higher than the normal healthy population, and some patients with insomnia may be more prone to stroke risk [7].As increasing studies showed that insomnia has a bidirectional relationship with stroke, which may be an independent risk factor for stroke.Further, stroke may also be a predisposing factor for insomnia [8].Therefore, it is essential to understand the relationship between stroke and post-stroke insomnia in an early stage to provide a clinical basis for the early prevention and treatment of post-stroke insomnia.The study aimed to investigate the prevalence of insomnia in patients with stroke.

Research design and method
The study was conducted and designed in strict accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [9,10].

Data source and selection process
Literature related to the occurrence of developmental insomnia in stroke patients was collected through PubMed, The Cochrane Library, Web of Science, and Embase databases until June 2023.

Search strategy
We searched the related literature by the subject terms, such as "Stroke", "Cerebrovascular Accident", "Insomnia", "Insomnia Disorder", etc.The following search strategy for the PubMed database (Fig 1).

Eligibility criteria and study selection
In the study, we included the cohort studies and cross-sectional studies about stroke patients who developed insomnia in English language.Stroke patients met the diagnostic criteria of the Essentials of Diagnosis of Various Cerebrovascular Diseases [11].Insomnia patients were diagnosed through recognized assessment tools such as the Pittsburgh Sleep Quality Index (PSQI), Hamilton Depression Scale (HDS), or self-reported symptoms of insomnia and met the diagnostic criteria of the American Academy of Sleep in 2014 [12].We excluded the duplicate records, case reports, reviews and so on.

Exclusion criteria
We excluded the duplicate literature, case reports, reviews and the the literature with incomplete data indicators, or the information was not available.

Data extraction
2.5.1 Literature screening and information extraction.LAT and TQJ screened the included literatures.The extracted information mainly included the basic information of the literatures: first author name, the time of publication, sample size, the country, the follow-up time and the number of positive cases.In case of disagreement between two researchers in the literature screening or data extraction process, the decision was submitted to the third researcher (YJW).
2.5.2Literature quality assessment.The methodological quality of the included studies was assessed using the Critical Appraisal Tool for Prevalence Studies [13,14].Any disagreements by the researchers were submitted to a third researcher (YJW).

Statistical analysis
In the study, we used systematic Meta-Analysis software version 3 to calculate the statistical analyse [15].The fixed effects model was used in P�0.10 and I 2 �50%, and random-effects model was used in P<0.10 and/or I 2 >50%, which was necessary to find the source of heterogeneity and perform subgroup analysis or sensitivity analysis [16][17][18].

Literature search results
We got 1507 literatures from databases, of which 469 were duplicates and hence excluded.Further, we excluded 927 studies by the exclusion criteria.Overall, 111 studies were retained for the full-text evaluation, and finally 26 studies were included in the meta-analysis (Fig 2) [7,.

Basic characteristics of the included studies
The 26 included studies (13 prospective cohort studies, 10 cross-sectional studies, 2 retrospective studies, and 1 multicenter observational study) were published between 2002 and 2023.Overall, the 26 studies included had 1,193,659 participants, of which 497,124 were patients with stroke.The details are shown in Table 1.

Quality of included studies
Table 2 shows the quality assessment of the included studies.69.23% (eighteen studies) studies determined the prevalence of insomnia in stroke patients in a sufficient sample size.To assess insomnia, most studies used standardized instruments or validated diagnostic criteria (80.77%).The details are shown in Table 2.

Meta-analysis
We used the random-effects model to pool prevalence of insomnia in patients with stroke.150,181 patients with stroke developed insomnia during the follow-up and the pool prevalence was 46.98% (95% CI: 36.91-57.18)(Fig 3).Moreover, nine studies examined the occurrence of insomnia in patients with IS or TIA.The result showed that the prevalence of insomnia among patients with IS or TIA was 47.21% (95% CI: 34.26-60.36)(Fig 4).
Four studies explicitly examined the prevalence of insomnia among IS or hemorrhage patients and the prevalence was 44.09% (95% CI: 19.84-69.92),while twelve studies did not specify the type of stroke (Fig 5).
In the subgroup analysis, we found that the incidence of insomnia was significantly higher in the patients with stroke at a mean age of �65 than patients with stroke at a mean age of <65 years, which was [47.18% (95% CI: 26.7-68.16)vs 40.50% (95% CI: 26.21-55.66),P<0.05] (Figs 8 and 9).
Moreover, concerning the follow-up duration of the participants, we found that the prevalence of insomnia was significantly higher in the follow-up duration was �3 years than those with a follow-up period <3 years (58.06% vs 43.83%, P < 0.001) (Figs 10 and 11).
In the end, the subgroup analyse was performed based on the use of insomnia assessment diagnostic tools (clinical assessment diagnostic tools vs self-report).Twenty-one studies used insomnia assessment diagnostic tools, and the insomnia rate in stroke patients was 49.31% (95% CI: 38.59-60.06)(Fig 12).Five studies used self-report, and the results indicated that the insomnia rate in stroke patients was 37.58% (95% CI: 13.44-65.63)(Fig 13).

Key findings
This study was an updated review about the prevalence of insomnia among patients with stroke.Further, 26 studies from 11 countries were included, of which 15 studies were conducted in Asia (57.69%) and the remaining studies were conducted outside Asia.Of the 26 included studies, 21 used diagnostic tools and 5 used nondiagnostic tools for assessing insomnia.Overall, our meta-analysis indicated that the rate of insomnia after stroke was 48.37%.It was estimated that incidence of IS or TIA (47.21%) was higher than that of unclassified stroke (41.51%); the rate of acute-phase stroke was higher (59.16%) than that of nonacute-phase stroke (36.31%); the proportion of patients with a mean age �65 years was higher (47.18%) than the proportion of those with a mean age <65 years (44.43%); the duration of follow-up �3 years (58.06%) was higher than the duration of follow-up <3 years (43.83%); and the rate of using a diagnostic tool for insomnia assessment was higher (51.16%) than the rate of using a nondiagnostic tool (37.58%).This suggested that post-stroke insomnia was a substantial global public health problem in patients with stroke who needed urgent attention for prevention and treatment.

Comparisons of the study findings with the available evidence
Our study found that the rate of insomnia after stroke (48.37%) was 1.27 times higher compared with the prevalence in the meta-analysis by Baylan et al. in 2019 (38.2%) [44].It indicated that the prevalence of post-stroke insomnia continued to increase yearly, and insomnia had a significant negative impact on patients.The data in this study indicated that sleep-related apnea was significantly associated with stroke, and obstructive apnea syndrome might increase the risk of stroke twice [1].A 4-year follow-up study in Taiwan, China revealed that compared with patients without insomnia, the incidence of stroke was significantly higher in insomnia patients [45].A similar meta-analysis showed that sleep duration was also associated with the risk of stroke, with a 5%-7% increase in stroke risk for every 1-h decrease in short sleep duration (RR = 1.05-1.07,95% CI: 1.01-1.12)[46,47].
Insomnia after stroke is associated with the acute or chronic phase of stroke.In this study, we found that the rate of insomnia was higher in the acute phase of stroke (59.16%) than in the nonacute phase of stroke (36.31%).Luisa et al. found that polysomnography in acute IS patients showed poorer sleep quality was associated with sleep efficiency, sleep-onset awakening time in stroke patients [48].Several factors usually caused insomnia in patients with stroke.Insomnia in patients with acute stroke was found to be associated with an increased risk of post-stroke psychiatric disorders [49].
Moreover, the age of patients with stroke and the duration of follow-up are also important factors influencing the rate of insomnia in patients with stroke.In the general population, insomnia may increase with age [50].Studies showed a significantly higher prevalence of insomnia in elderly people [51].Nick Glozier et al. found that the prevalence of insomnia was 16% after 6 months of stroke and 23% after 12 months of stroke [38].The aforementioned study suggested that older patients with stroke might have an increased likelihood of experiencing insomnia during the follow-up period, and this likelihood seems to grow over time.
Insomnia assessment and diagnostic tool is also one of the factors affecting the rate of insomnia.This study found that the prevalence of insomnia using the Insomnia Assessment Diagnostic Tool was 51.16%, which was higher than the prevalence of self-reported insomnia (37.58%).In contrast, in study using the insomnia assessment and diagnostic tool, the prevalence of insomnia was different in acute phase and subacute phase stroke (32.5% vs 34.8%), whereas the overall prevalence of self-assessed insomnia also was different in acute phase and subacute phase stroke (47.1% vs 50.4%) [52].Further large-sample studies are needed to validate these findings.
This study had some limitations.First, the study quality was not an exclusion criterion, which might have contributed to the differences in the prevalence of insomnia after stroke.Studies used different tools for assessing and diagnosing insomnia, which might also have led to biased conclusions.Second, we did not study the treatment of patients with stroke and its effect on the development of insomnia.

Conclusions
Stroke may be a predisposing factor for insomnia.Insomnia is more likely to occur in acutephase stroke, and the prevalence of insomnia increases with patient age and follow-up.Further, the rate of insomnia is higher in patients with stroke who use the Insomnia Assessment Diagnostic Tool.

Table 2 .
(Continued) Data analysis conducted with sufficient coverage of the identified sample.Q6.Valid methods used for the identification of insomnia or insomnia symptoms.Q7.Valid methods used for the identification of stroke.Q8.Condition measured in a standard, reliable way for all participants.