Gender‐specific estimates of sleep problems during the COVID‐19 pandemic: Systematic review and meta‐analysis

Summary The outbreak of the novel coronavirus disease 2019 (COVID‐19) changed lifestyles worldwide and subsequently induced individuals’ sleep problems. Sleep problems have been demonstrated by scattered evidence among the current literature on COVID‐19; however, little is known regarding the synthesised prevalence of sleep problems (i.e. insomnia symptoms and poor sleep quality) for males and females separately. The present systematic review and meta‐analysis aimed to answer the important question regarding prevalence of sleep problems during the COVID‐19 outbreak period between genders. Using the Preferred Reporting Items for Systematic Reviews and Meta‐Analyses guideline and Newcastle–Ottawa Scale checklist, relevant studies with satisfactory methodological quality searched for in five academic databases (Scopus, PubMed Central, ProQuest, Web of Science , and EMBASE) were included and analysed. The protocol of the project was registered in the International Prospective Register of Systematic Reviews (PROSPERO; identification code CRD42020181644). A total of 54 papers (N = 67,722) in the female subgroup and 45 papers (N = 45,718) in the male subgroup were pooled in the meta‐analysis. The corrected pooled estimated prevalence of sleep problems was 24% (95% confidence interval [CI] 19%–29%) for female participants and 27% (95% CI 24%–30%) for male participants. Although in both gender subgroups, patients with COVID‐19, health professionals and general population showed the highest prevalence of sleep problems, it did not reach statistical significance. Based on multivariable meta‐regression, both gender groups had higher prevalence of sleep problems during the lockdown period. Therefore, healthcare providers should pay attention to the sleep problems and take appropriate preventive action.


| INTRODUC TI ON
The outbreak of the novel coronavirus disease 2019  changed most people's lifestyles globally. Indeed, many countries and governments used different policies (e.g. city lockdown, boarder control, online teaching, and special distancing) to slow down the COVID-19 infection rate Chen, Chen et al., 2021); as COVID-19 was found to have an extraordinary transmission rate and cause an alarming number of deaths (Ahorsu, Lin, Imani et al., 2020;Mamun et al., 2021). With the high prevalence and level of mortality (WHO, 2020), COVID-19 has impacted peoples psychological health. Indeed, numerous studies have found that  together with the reactions toward controlling COVID-19 infection are associated with different aspects of psychological health, including depression, anxiety, stress, and sleep problems (Ahorsu, Chang et al., 2020;Lin, Broström et al., 2020, Lin, Imani et al., 2020. Among the psychological health aspects, sleep is one of the major concerns for healthcare providers  for the following reasons. First, sleep is an essential component for individuals having effective cognitive and emotional processing, and a good night's sleep is proposed to be vital for all people (Garbarino et al., 2016;Kopasz et al., 2010;Tarokh et al., 2016;Yaffe et al., 2014). Second, ample evidence has shown that sleep is a key factor for individuals maintaining satisfactory and good health, including physical functioning, mental functioning, social functioning, spiritual functioning, and overall quality of life (Garbarino et al., 2016;Gradisar et al., 2008;Shochat et al., 2014). Third, an association between good sleep and health behaviours have been proposed . However, individuals living in the modern world have different obstacles for achieving good sleep , given that the technology today contributes to sleep disturbance (Alimoradi et al., 2019). Moreover, recent research shows that problematic social media use, a behaviour found to have increased during the COVID-19 outbreak (Hashemi et al., 2020;Lin, Broström et al., 2020), is associated with poor sleep (Wong et al., 2020). In short, there is a need to investigate in-depth the sleep problems occurring during the COVID-19 outbreak period.
The available literature on COVID-19 shows the findings of sleep problems.  studied sleep problems amongst healthcare workers and found different prevalence rates of insomnia between non-medical healthcare workers (e.g. volunteers in the hospital, medical students, and community workers; prevalence of 38.4%) and medical healthcare workers (e.g. medical doctors and nurses; prevalence of 30.5%). Wang, Song et al. (2020) also examined sleep problems in four populations and found different prevalence rates as well. The prevalence of sleep problems among medical staff was 66.1%, in non-medical staff was 47.8%, in frontline healthcare providers was 68.1%, and in non-frontline healthcare providers was 64.5%. Although the information on sleep problems during the COVID-19 outbreak period has been studied and reported, healthcare providers need synthesised information regarding sleep problems across gender. However, to the best of the present authors' knowledge, no empirical studies have focussed on the sleep problems between genders during the COVID-19 pandemic, although the studies have controlled for gender in their statistical analyses.
Gender is an important issue for sleep because different treatments may be designed or used for different genders. More specifically, prior evidence has shown that males and females have different processes in brain functions (Xin et al., 2019). Therefore, males and females may not always share the same values on everything. For example, prior research indicates that males as compared with females appreciate physical activity more (Ou et al., 2017).
Additionally, males and females report different levels of psychological health (including quality of life) from children and older people (Lin et al., 2016;Su et al., 2013). Therefore, it is important for healthcare providers to understand sleep problems separately for males and females during the COVID-19 outbreak period.
To answer the important question regarding prevalence of sleep problems during the COVID-19 outbreak period across gender, the present study was designed and conducted as a systematic review and meta-analysis. With the robust methods used in the present review, information on sleep problems across gender were synthesised and should assist healthcare providers in understanding the impacts of the COVID-19 outbreak on sleep.

| ME THODS
This systematic review is reported based on the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guideline (Moher et al., 2010), a systematic literature search was done in five academic databases, relevant studies were abstracted, and their methodological quality was assessed using the Newcastle-Ottawa Scale (NOS) checklist. Findings were synthesised using a metaanalysis approach. Results of the present paper are part of the findings from a larger project, the protocol of this project was registered in the International Prospective Register of Systematic Reviews (PROSPERO; identification code CRD42020181644) (Alimoradi & Pakpour, 2020).

| Search strategy
Five academic databases including Scopus, PubMed Central, ProQuest, Web of Science (WoS), and the Excerpta Medica data-BASE (EMBASE) were searched systematically. The search terms were extracted from published reviews and primary studies besides PubMed Medical Subject Headings (MeSH). Specifically, the Patientproblem, Exposure, Comparison, and Outcome (PECO) framework was used to determine search terms. In this regard, the "patientproblem" was any human population, the "exposure" was COVID-19 pandemic with a variety of factors contributing to sleep problems (including stress, reduced light exposure, extended working hours, and changed lifestyle), the "comparison" was none given that all the populations were impacted by exposure to the COVID-19 pandemic, and the "outcome" was sleep. The main search terms were sleep and COVID-19. The Boolean search method (AND/OR/NOT) was used to develop the search query. Search syntax was customised based on the advanced search attributes of each database. The search strategy is provided as Additional File 1. Additionally, reference lists of included studies were searched to increase the likelihood of retrieving relevant empirical studies.

| Inclusion criteria
Observational studies, including case-control and cross-sectional studies, were included if relevant data relationships were reported.
English, peer-reviewed papers published between December 2019 and February 2021 were included. However, the papers were further screened to ensure that the data collection period was during the COVID-19 pandemic or COVID-19 endemic in mainland China.
No limitation was imposed regarding participants characteristics.

| Primary outcome
Gender-specific estimation of sleep problems prevalence during the COVID-19 pandemic was the primary outcome.
2.2.2 | Secondary outcomes 1. Assessing the heterogeneity and its possible sources.
2. Influencing variables (e.g. age and marital status) in gender-specific sleep problems prevalence during the COVID-19 pandemic.

| Study screening and selection
In the first step, the title and abstract of all retrieved papers were screened based on the inclusion criteria. The full texts of potentially relevant studies were further examined based on the aforementioned criteria. In this process, relevant studies were selected.

| Quality assessment
The NOS was used to evaluate the methodological quality of the studies in observational studies. Three characteristics of selection, comparability, and outcome are examined with the NOS checklist. The checklist has three versions for evaluating cross-sectional studies (seven items), case-control (eight items), and cohort (eight items). Despite a slight difference in number and content of items, each item is rated with a star, except the comparability that can have two stars, thus resulting in a maximum score of 9. Studies with <5 points are classified as having a high risk of bias (Luchini et al., 2017). No studies were excluded based on the quality. But subgroup analysis was conducted to assess the impact of quality on pooled effect size.

| Data extraction
A pre-designed form was prepared to extract data from included studies. Data including first author's name, collection date, study design, country, number of participants, gender, mean age, scale used to assess sleep problems, numerical results regarding the frequency of sleep problems. In the process of data extraction, two Excel sheets were initially designed, with one summarising the features of the included studies (e.g. author name and publication year) and the other evaluating methodological quality. The required data from the articles were later entered into another Excel datasheet for coding and preparing for analysis using STATA statistical software.
It should be noted that study selection, quality assessment, and data extraction were processes performed independently by two reviewers. In whole processes (i.e. study selection, quality assessment, and data extraction) disagreements were resolved through discussion by two independent reviewers. A third party was not required to resolve disagreements between the two independent reviewers because there were only minor disagreements, and both reviewers easily reached a consensus.

| Data synthesis
A quantitative synthesis using STATA software version 14 was conducted. Meta-analysis was run using a random effect model, as it was proposed that included studies were taken from different populations both within-and between-study variances should be accounted for (Hox & Leeuw, 2003). The Q Cochrane statistic was used to assess heterogeneity. Also, the severity of heterogeneity was estimated using the I 2 index. Heterogeneity is interpreted as mild when I 2 is <25% and is considered moderate when I 2 is 25%-50%, and severe heterogeneity is diagnosed when I 2 is 50%-75%.
Prevalence of sleep problems was the selected key measure for the present study. This pooled estimate of this key measure with 95% confidence interval (CI) is reported. Subgroup analysis or meta-regression was done to find possible sources of heterogeneity and influencing variables on gender-specific sleep problems prevalence. Funnel plot and the Begg's test were used to assess publication bias (Rothstein et al., 2005). Potential publication bias was corrected with the "fill-and-trim" method (Duval & Tweedie, 2000).

| Study screening and selection process
The initial search of the five databases resulted in 7,263 studies: Scopus (n = 2,518), WoS (n = 474), PubMed (n = 338), EMBASE (n = 1,426), and ProQuest (n = 2,507). After removing duplicate papers, a further 5,647 papers were screened based on title and abstract. Finally, 555 papers appeared to be potentially eligible and their full texts were reviewed. In this process, 54 studies in the female subgroup and 45 studies in the male subgroup met the eligibility criteria and were pooled in the meta-analysis. Figure 1 shows the search process based on the PRISMA flowchart.  (14), and patients with COVID-19 (seven). Most of the studies were cross-sectional (43 studies). The two remaining studies had a longitudinal design and collected data during the COVID-19 pandemic and baseline data were extracted. The ISI and PSQI were used to assess sleep problems (in 25 and 14 studies, respectively). Considering NOS >5 as high quality, 71% of the included studies (32 papers) were categorised as high-quality.

| Estimation of sleep problem prevalence
The pooled estimated prevalence of sleep problems was 31% (95% CI 28%-35%; I 2 : 97.58%, tau 2 : 0.01). Figure 2 provides a Forest plot of the pooled prevalence of sleep problems in this group.
Subgroup analysis (Table 2) showed that the prevalence of sleep problems was higher in longitudinal versus cross-sectional studies (48% versus 31%). Although prevalence of sleep problems appeared to be different among male healthcare professionals (34%), the general population (29%) and patients with COVID-19 (39%), these differences were not statistically significant considering overlap in the 95% CI of pooled prevalence among these groups (26%-43% for healthcare professionals, 24%-33% for general population, and 27%-50% for patients with COVID-19). Based on multivariable meta-regression (Table 4), being in lockdown period, quality of studies, and measure used to assess sleep problems were significant These variables together explained 100% of the variance.
Begg's test (p = 0.006) and funnel plot ( Figure 3) consider probability of publication bias. Meta trim was used to correct publication bias. Based on the trim method, eight studies were imputed, and the corrected prevalence of sleep problems was 27% (95% CI 24%-30%). The corrected funnel plot is provided in Figure 4. Also, sensitivity analysis showed that pooled effect size was not affected by the effect of a single study.  (15), and patients with COVID-19 (seven). Most of the studies were cross-sectional (52 studies). The two remaining studies had a longitudinal design and collected data during the COVID-19 pandemic and baseline data were extracted. The ISI and PSQI were used to assess sleep problems (in 31 and 17 studies, respectively). Considering NOS >5 as high quality, 70% of the included studies (38 papers) were categorised as high-quality. Table 1 provides the summary characteristics of the included studies.

| Estimation of sleep problem prevalence
The pooled estimated prevalence of sleep problems was 41% (95% CI 36%-45%; I 2 : 99.43%, tau 2 : 0.03). Figure 5 provides a Forest plot regarding the pooled prevalence of sleep problems in this group.
Subgroup analysis (Table 2) showed that the prevalence of sleep problems was higher in longitudinal versus cross sectional studies (55% versus 41%). Although prevalence of sleep problems appeared to be different among female patients with COVID-19 (51%), healthcare professionals (41%), and the general population (38%), these differences were not significantly different considering the overlap in 95% CI of pooled prevalence among these groups (31%-51% for healthcare professionals, 32%-44% for general population, and 42%-60% for patients with COVID-19). Based on univariate metaregression (Table 3), country and percentage of married participants were other significant predictors of sleep problems prevalence among women. In multivariable meta-regression (Table 4) being in lockdown and study quality were significant predictors of sleep problems prevalence among female participants, which explained 34.18% of the variance.
As indicated above, the Begg's test (p = 0.08) and funnel plot ( Figure 6) consider probability of publication bias. Meta trim was used to correct publication bias. Based on the trim method, 22 studies were imputed, and the corrected prevalence of sleep problems was 24% (95% CI 19%-29%). The corrected funnel plot is provided in Figure 7. Also, sensitivity analysis showed that the pooled effect size was not affected by the effect of a single study.

| DISCUSS ION
The present systematic review and meta-analysis aimed to provide timely information for healthcare providers to understand how the  (Cénat et al., 2020). Notwithstanding, female health professionals appear to be more likely to experience sleep problems compared to their counterparts in the general population, but such differences did not emerge in men.
As indicated, most of the data retrieved for the present systematic review and meta-analysis originated from cross-sectional designed studies. Notwithstanding, we surmise that the fear and stress associated with COVID-19 may be one of the major reasons contributing to the high prevalence of sleep problems. More specifically, social media and news channels have continuously routinely reported on daily deaths and on the number of cumulative infected cases of COVID-19 both at the national and global scales, and such intensive media exposure is likely to generate the anxiety and stress

TA B L E 1 (Continued)
F I G U R E 2 Forest plot for the pooled prevalence of sleep problems in the male group. CI, confidence interval; ES, effect size that facilitate the emergence of sleep problems , Lin, Imani et al., 2020. Indeed, higher levels of psychological distress and signs of mental disorders have been reported during this pandemic among different populations worldwide (Mamun et al., 2021;Rodríguez-Rey et al., 2020;Wang, Pan et al., 2020) and significant sleep difficulties have been identified in the context of major public health threats (e.g. Ebola) (Cates et al., 2018;Lehmann et al., 2015).
The reasons for the higher prevalence of sleep problems in females are unclear, but possibly may reside in the underlying brain structural differences across sexes (Xin et al., 2019). Therefore, exposure to the same circumstances may yield different perceptions and lead to divergent emotional processing. Indeed, prior evidence found that self-reported outcomes on subjective health (e.g. quality of life) differ between males and females (Lin et al., 2016;Su et al., 2013). Additionally, women are more likely to report psychological problems in response to taxing situational settings (Wang et al., 2017). Finally, issues such as insomnia exhibit clear gender dimorphic features (Kocevska et al., 2020;Silva-Costa et al., 2020;Sivertsen et al., 2021).
The sleep problems among healthcare professionals found in the present systematic review and meta-analysis could be attributed to the interactions between the COVID-19 pandemic and the specific attributes of the jobs. From the perspective of the COVID-19 pandemic, health professionals, especially those who had to be in direct contact with patients with COVID-19 and those who were at high risk of being exposed to the COVID-19 virus, had higher levels of worry and psychological distress. The higher levels of worry and psychological distress are likely to subsequently foster the development of their sleep problems (Fidanci, derinöz Güleryüz, & Fidanci, 2020). From the perspective of the job itself, health professionals, especially those who work in a large hospital, have irregular work   (Chang et al., 2020;Chen, Chen et al., 2021, Chen, Wang et al., 2021Lin, Broström et al., 2020, Lin, Imani 2020Mamun et al., 2021;Pramukti et al., 2020) and such measures could affect the prevalence rates of sleep problems.
In summary, a relatively high prevalence of sleep problems

AUTH O R CO NTR I B UTI O N S
Each author made a substantial contribution to project design, data collection or data analysis. Additionally, all authors contributed to the preparation of this manuscript.

CO N FLI C T O F I NTE R E S T
All authors have no conflicts to declare.

DATA AVA I L A B I L I T Y S TAT E M E N T
Data sharing not applicable to this article as no datasets were generated or analysed during the current study.

FU N D I N G I N FO R M ATI O N
The open access was funded by Jönköping University.  (2020). Internet-related behaviors and psychological distress among schoolchildren during COVID-19 school suspension. Journal F I G U R E 7 Corrected funnel plot based on the fill-and-trim method in the male subgroup