Sleep patterns during the pandemic and covid-related stress: a large-scale study in the general population of Serbia and Republic of Srpska (Bosnia and Herzegovina)

Abstract

Background

During the COVID-19 pandemic many people experienced psycho-social stress which affected their sleep. This study was conducted during the third and fourth epidemic wave when international borders of Serbia and Bosnia-Herzegovina were entirely open, no curfew or wearing face masks were imposed, but the highest rates of COVID-19-related fatalities were reported. The aim of this study was to examine sleep patterns with COVID-related stress during the pandemic.

Methods

Anonymous paper questionnaires were distributed from September 2020 to October 2021 across 8 cities in Serbia and Republic of Srpska (Bosnia and Herzegovina). Socio-epidemiologic characteristics, the COVID Stress Scales-CSS and the Perceived Stress Scale were administered. Sleeping patterns before and during the COVID-19 pandemic (bedtime, time spent sleeping and sleep quality) were recorded.

Results

Responses of 2,301 participants suggested that bedtime after midnight before and during the pandemic did not differ (13.4% vs. 14.8%, respectively). Most participants reported similar length of sleep before and during the pandemic (around 7 h), although 11% of them reported that during the pandemic they slept more often compared to pre-pandemic sleeping schedule. There was an increase in prevalence of poor sleep quality during the pandemic (4.5% vs. 9.4%, respectively). Sleeping more often compared to pre-pandemic sleeping schedule and poor sleep quality during the pandemic were independently associated with a higher CSS.

Conclusion

Proportion of people who reported poor sleep quality doubled in the pandemic. Optimizing sleep quality in crises among people who experience poor sleep quality should be prioritized when managing public health emergencies.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12963-026-00485-2.

Introduction

To maintain good physical and mental health experts recommend 7 to 9 h of sleep per night to adults and 7 to 8 h of sleep to older adults [1]. There are various environmental, social and behavioral influences that affect duration and quality of sleep [2]. As a result, in different circumstances, people might not be able to achieve the recommended sleep hours. For example, a major event that fundamentally changed day-to-day lives of people worldwide was the COVID-19 pandemic. Expectedly, many people experienced psycho-social stress because of fear of catching and dying from the infection, maintaining livelihood and being overwhelmed with pandemic-related information [3, 4]. All these factors combined have had a major role in the global sleep disturbance [5].

In Serbia and Republic of Srpska (Bosnia and Herzegovina), first cases of COVID-19 were identified at the beginning of March 2020. At the end of March 2020 the lockdown was imposed in both countries [6]. Borders were closed, and the curfew was on from 5 pm to 5 am from Monday to Friday, as well as throughout the weekend, from Friday 5 pm to Monday 5 am. Older people (aged above 65 years) were not allowed to leave their homes. After the lockdown had ended in May 2020, a second epidemic wave was recorded over the summer months and the third epidemic wave began in November 2020 lasting until the early spring 2021. The third epidemic wave saw the highest case-fatality rate throughout the pandemic associated with the predominant Delta variant. The fourth wave occurred in the summer of 2021 and the fifth wave began at the end of 2021, where the Omicron variant was prevalent [7]. Throughout the COVID-19 pandemic in Serbia and Republic of Srpska (Bosnia and Herzegovina), face masks were recommended, but not enforced. Vaccines from four manufacturers were available [8] and cash reward was offered to all people who received COVID-19 vaccines [9]. However, COVID certificates were mandated in cafes and restaurants.

Social isolation and loneliness, working remotely or not being able to do so due to the nature of work, organization of family and (multigenerational) household, anxiety in anticipation of potential symptoms and severe clinical course of COVID-19 and possible economic losses, painted a complex landscape that many people were required to navigate [10]. Changes in sleep patterns that resulted from adjusting to the new pandemic living conditions could have had prolonged impact on COVID-related stress levels. Nevertheless, thus far, sleep behavior during the pandemic and its association with stress levels have not been explored in the general population of Serbia and Bosnia and Herzegovina. This study was conducted over the period when international borders were entirely open, no curfew and face masks were imposed, however, at the same time the highest rates of COVID-19-related fatalities were reported. This social and environmental setting could have precipitated a rise in sleep disturbance and stress during the pandemic. Data such as these may be helpful to better understand and mitigate sleep disturbances in future health crises.

The purpose of this study was to examine sleep patterns with COVID-related stress during the pandemic in the general Serbian-speaking population.

Methods

Study settings

The present study was conducted in 8 randomly chosen cities of two regions/countries where Serbian language is spoken: Republic of Serbia and Republic of Srpska – Bosnia and Herzegovina throughout a one-year period (September 2020 to October 2021), when the third and the fourth epidemic waves occurred. This period was also the time when the highest COVID-19 case-fatality rates in Serbia and Bosnia and Herzegovina. No curfew or lockdown were in place at the time of survey. In fact, throughout this period, borders were open in both Serbia and Bosnia-Herzegovina and international travel was entirely functional. Nevertheless, data collection timing included the period of lower COVID-19 occurrence in between the epidemic waves, which could have influenced sleeping patterns.

To select the cities for participant recruitment, we printed the names of all official administrative cities in both regions/countries (29 in Serbia and 6 in Republic of Srpska – Bosnia and Herzegovina) on separate papers and placed them in a non-transparent container. In a table of random numbers, we hit number 4 and then draw from the container every fourth paper with names of cities. However, because smaller cities had just one and larger had multiple municipalities, to overcome the discrepancies in the city sizes, we decided to distribute the questionnaires in only one randomly chosen municipality per selected city. We printed the names of all municipalities on separate papers and draw them out from a container.

Selection of study participants

People who came to the chosen municipality office headquarters, to engage in regular administrative business, were invited to participate in the study. In Serbia and Republic of Srpska – Bosnia and Herzegovina, civil administration affairs (taxation, legal counseling, certificate issuance etc.) are still mostly managed in person. Therefore, almost all citizens have had to visit their municipality office at least once in their life. Nevertheless, options to complete forms digitally were introduced as of recently. For this reason, some people must have opted to review and request forms online, even though most people were still visiting municipality headquarters in person.

The inclusion criteria were: being ≥ 18 years and speaking Serbian language fluently. The exclusion criteria were: reporting psychiatric disorders previously diagnosed by a physician (F-coded diagnoses of the International Classification of Diseases 10th revision) and providing less than 90% of answers.

We approached 2,907 persons out of which we had to exclude 38 individuals due to confirmed psychiatric disorders, 154 due to not responding to 90% of the items in the questionnaire, while 414 refused participation. The response rate was 85.7%.

Instruments

Data were collected using anonymous questionnaires. All the applied instruments were self-administered in a paper-and-pencil form. At any moment, at least one researcher was present to provide all necessary explanations. The questionnaire set included a general socio-demographic and behavioral questionnaire, the COVID Stress Scales, the Perceived Stress Scale and the COVID-19 attitudes questionnaire.

Outcome

The study outcome was the score on the COVID Stress Scales (CSS). This questionnaire was designed to better understand and assess COVID-19-related distress and health-related anxiety during the pandemic [11]. The scale has 36 items classified into six domains: danger (DAN), socio-economic consequences (SEC), xenophobia (XEN), contamination (CON), traumatic stress symptoms (TSS) and compulsive checking and reassurance seeking (CHE). Items are rated regarding frequency or intensity from 0 (never/not at all) to 4 (almost always/extremely). The scores for each of the six domains are calculated as the sum of ratings for each item in that domain. Higher scores indicate more intense or more frequent perceptions. The total CSS score is the sum of all six domain scores and ranges from 0 (low COVID-19-related distress) to 144 (severe COVID-19-related distress). The CSS has previously been translated and validated in the Serbian population [12].

Exposures

The study outcomes were the following:

•  bedtime at the time of survey i.e. during the pandemic (evaluated by a question “At what time do you usually go to sleep now?” A dash was left for the participant to write in the time).

•  bedtime before the pandemic (evaluated by a question “At what time did you usually go to sleep before the COVID-19 pandemic began?” A dash was left for participant to write in the time).

•  sleep duration at the time of survey (evaluated by a question “On average, how long do you sleep during the night now? A dash was left for participant to write in number of hours spent sleeping).

•  change in sleep schedule compared to pre-pandemic sleeping schedule (evaluated by a question “Since the beginning of the COVID-19 pandemic do you sleep: a) longer/b) more often (including daytime napping)/c) shorter/d) the same”. Participants were asked to encircle the option that refers to their experiences; sleeping more often referred to daytime naps in addition to sleeping at night)

•  sleep quality (evaluated by a question “How do you rate your sleep quality since the beginning of the COVID-19 pandemic? a) very poor/b) poor/c) average/d) good/e) very good”. Participants were asked to encircle the option that reflects their experiences).

Covariates

Socio-demographic characteristics (gender, age, education level, being partnered, employment status before and during the pandemic, income before and during the pandemic, weight change during the pandemic, money loss during the pandemic), lifestyle habits and health behaviors (tobacco smoking before and during the pandemic, alcohol intake before and during the pandemic, using immunity boosters such as vitamins and herbal supplements), information regarding contact people with confirmed or suspicious COVID-19, history of COVID-19 in participants and their family, having had COVID-19 symptoms, getting information regarding COVID-19 from medical (health care providers, medical literature) and non-medical sources of information.

The Attitude toward COVID-19 scale (AT scale) was developed by the authors for the purpose of this study. It contains 15 items exploring participants’ worry about catching COVID-19 as well as actions they take (i.e. behaviors they practice) to lower the risk of catching COVID-19. Responses were scored on a 5-point Likert scale (1 strongly disagree to 5 strongly agree). Total score was obtained by adding item-level scores and it ranged from 15 to 75. Higher total score indicated increased level of COVID-19 related worry. Psychometric testing of the AT scale suggested a good internal consistency as per the Cronbach’s alpha coefficient of 0.702. The AT score was included in the analysis to account for feelings of worry that were not covered by the CSS.

The Perceived Stress Scale (PSS) is the most frequently used scale to measure perceived stress relative to current life conditions and expectations for future [13]. It assesses the degree to which individuals evaluate their life circumstances and situations as stressful as well as perception of unpredictability, uncontrollability, and sense of overload in life. The PSS consisted of 10 items rated from 0 (never) to 4 (very often). The PSS scores are obtained by reversing responses to positive items (items 4, 5, 7 and 8) and then adding all item-level scores. Higher total scores indicate greater levels of stress. The PSS has previously been translated and validated in the Serbian population [14]. The PSS was included in the analysis to account for unexpected stress unrelated to COVID-19 itself e.g. work-related stress or personal stress associated with separating with partner or partner/family being sick or hospitalized.

Statistical analysis

To describe the study sample, we used the mean value and standard deviation (SD) for demographic characteristics and the CSS and PSS scores when showing continuous variables. Frequencies and proportions were used to describe categorical characteristics. Differences in the investigated parameters were tested by the Kruskal-Wallis χ² test. We used the Spearman’s rank correlation to investigate the relationships between the tested parameters.

The multiple linear regression model was applied to identify factors associated with a higher total CSS score. First, the univariate (unadjusted) linear regression model was tested. In this model, the dependent variable was the continuous total CSS score. The independent variables were sleeping patterns as reported by the study participants – bedtime before the pandemic (continuous variable), bedtime during the pandemic (continuous variable), sleep duration before the pandemic (continuous variable in hours), sleep duration in the pandemic (continuous variable in hours), sleep schedule compared to pre-pandemic schedule (0-shorter sleep duration compared to pre-pandemic, 1-unchanged duration of sleep, 2-longer sleep duration during the pandemic, 3-sleeping more often i.e. having daytime naps), sleep quality before the pandemic (from quite poor to very good) and sleep quality during the pandemic (from quite poor to very good).

In the multiple linear regression model, the dependent variable was the continuous total CSS score. In addition to sleeping patterns described in the univariate model, the independent variables included the AT (continuous variable) and PSS scores (continuous variable), gender (1-male, 2-female), age (continuous variable), education level (years of schooling as a continuous variable) having chronic illnesses (1-yes, 0-no), weight change during the pandemic (1-yes, 0-no), being in a relationship (1-partnered, married or cohabitating, 0-single, divorced, widowed), employment status before (1-yes, 0-no) and during the pandemic (1-yes, 0-no), income (continuous variable), money loss during the pandemic (0-no, 1-unchanged, 2-gains), smoking before (1-yes, 0-no) and during the pandemic (0-less, 1-same, 2-more compared to pre-pandemic), alcohol intake before (1-yes, 0-no) and during the pandemic (0-less, 1-same, 2-more compared to pre-pandemic), contact with people who had confirmed (0-no, 1-unaware, 2-yes) or suspicious COVID-19 (0-no, 1-unaware, 2-yes), history of COVID-19 (1-yes, 0-no), having family members with COVID-19 (1-yes, 0-no), having COVID-19 symptoms (1-yes, 0-no), using immunity boosters (1-yes, 0-no) and receiving information on COVID-19 (0-does not follow information on COVID-19, 1-from non-medical sources, 2- from medical sources, 3-from all sources).

For all the models we evaluated probability level and adjusted R². Higher values of R² and adjusted R² indicated better model fit (R² ≥0.75 substantial, 0.25 to 0.75 moderate, < 0.25 weak). The adjusted R² is always lower than R², but if it is significantly decreased, it suggests that added predictors did not improve the models fit. The probability of p < 0.05 were accepted as statistically significant. Moreover, we performed homoscedasticity analysis of residuals and multicollinearity diagnostics (variance inflating factor - VIF values assessment) to confirm the model fit adequacy. The VIF of < 3 indicated a low correlation among the variables, while the model should not include variables with the VIF of ≤ 5. Analyses were performed using the Statistical Package for Social Sciences (SPSS) statistical software for Windows version 22 (SPSS, Inc, Chicago, IL).

Results

Study sample

This study included 2,301 participants out of which 54.9% were female (p = 0.001). Description of study participants is presented in Tables 1 and 2.

Table 1.

Age, income, COVID-19 stress scale scores, COVID-19 attitude score, Perceived stress score and sleep duration in the study sample

Parameters	Mean	Standard deviation	Median	Interquartile range (25–75%)
Age	36.71	13.82	35.0	26.0–48.0
Income (Euro)	640.61	327.39	583.33	416.7–791.7.7.7
Attitude score	43.94	10.87	44.0	37.0–51.0
Perceived Stress Scale score	19.43	5.05	19.0	16.0–22.0
Total COVID Stress Scale - CSS	32.73	23.87	28.0	15.2–46.0
1-Danger	32.73	23.87	10.0	6.0–14.0
2-Socio-economic consequences	9.72	5.37	2.0	0.0–6.0
3-Xenophobia	4.18	5.41	5.0	1.0–11.0
4-Contamination	6.51	6.21	3.0	1.0–8.0
5-Traumatic stress symptoms	4.98	5.15	1.0	0.0–5.0
6-Compulsive checking and reassurance seeking	2.91	4.31	3.0	0.0–7.0
Sleep duration before pandemic	7.36	1.13	7.0	7.0–8.0
Sleep duration in pandemic	7.33	1.22	7.0	7.0–8.0

Table 2.

Demographic characteristics and health behaviors in the study sample

Parameters		Frequency	Percent	p
Gender	male	1037	45.1	0.001
	female	1264	54.9
Age group	< 25	551	23.9	0.001
	25 to 50	1282	55.7
	≥ 50	468	20.3
Relationship	yes	1191	51.8	0.091
	no	1110	48.2
Education	8 years	120	5.2	0.001
	12 years	1138	49.5
	more	1043	45.3
Employment	no	782	34.0	0.001
	yes	1519	66.0
Chronic illnesses	no	1471	63.9	0.001
	yes one	516	22.4
	yes more than one	314	13.6
Smoking	no	1391	60.5	0.001
	yes	910	39.5
Alcohol drinking	no	694	30.2	0.001
	yes	1607	69.8

The majority of participants were 20 to 50 years old (age range from 18 to 90 years) and had no chronic illnesses. There were no differences between participants who were partnered and those who were not (single, widowed, other). Few participants (5.2%) finished primary school only, while secondary and university level of education were almost equally represented in the study sample. Most participants (66%) were employed throughout the COVID-19 pandemic. The mean income corresponded to the average salary in Serbia and Republic of Srpska. Study participants generally did not report having contact with COVID-19 positive people. Moreover, only 10.2% of participants had COVID-19. Study participants informed themselves about COVID-19 largely from non-medical sources (58.2% got informed from media, Internet, friends and family, etc.) (Table 3).

Table 3.

Experiences and behaviors during the COVID-19 pandemic in the study sample

Parameters		Frequency	Percent	p
CSS category	< 100	2259	98.2	0.001
	≥ 100	42	1.8
Money loss in pandemic	no	1745	75.8	0.001
	yes	556	24.2
Weight change in pandemic	loss	400	17.4	0.001
	same	1245	54.1
	gain	656	28.5
Smoking in pandemic	less	157	17.5	0.001
	same	569	63.4
	more	172	19.2
Alcohol use in pandemic	less	665	31.1	0.001
	same	1267	59.2
	more	209	9.8
Immunity boosters use	no	972	42.2	0.001
	herbal	386	16.8
	vitamins	943	41.0
COVID-19 contact	no	851	37.0	0.001
	do not know	631	27.4
	yes	819	35.6
Suspicious contact	no	798	34.7	0.363
	do not know	759	33.0
	yes	744	32.3
COVID-19 in family	no	1531	66.5	0.001
	yes	770	33.5
COVID-19 history	no	2067	89.8	0.001
	yes	234	10.2
COVID-19 symptoms	no	1394	60.6	0.001
	yes	907	39.4
Information	no info at all	102	4.4	0.001
	other sources	1339	58.2
	medical sources	70	3.0
	all sources	790	34.3

Sleep patterns

Significantly more study participants went to sleep before midnight, both before and during the pandemic. The prevalence of people who went to sleep after midnight before and during the pandemic did not differ (13.4% vs. 14.8%). Most participants also reported similar sleep duration before and during the pandemic (around 7 h), although 11% of them reported that during the pandemic they slept more often compared to pre-pandemic sleeping schedule. Most participants reported average or good quality of sleep before and during the pandemic. However, there was an increase in prevalence of study participants who reported poor sleep quality during the pandemic (Table 4).

Table 4.

Sleeping patterns before and during COVID-19 pandemic in the study sample

Parameters		Frequency	Percent	p
Bedtime before pandemic	≤midnight	1992	86.6	0.001
	> midnight	309	13.4
Bedtime in pandemic	≤ midnight	1960	85.2	0.001
	> midnight	341	14.8
Sleeping schedule in pandemic compared to pre-pandemic	same as before	1742	75.7	0.001
	shorter sleep duration	158	6.9
	longer sleep duration	147	6.4
	sleeping more often	254	11.0
Sleep quality before pandemic	quite poor	24	1.0	0.001
	poor	83	3.6
	average	785	34.1
	good	1011	43.9
	very good	398	17.3
Sleep quality in pandemic	quite poor	42	1.8	0.001
	poor	176	7.6
	average	849	36.9
	good	896	38.9
	very good	338	14.7

Stress associated with COVID-19

Mean ± SD Total CSS score was 32.73 ± 23.87 indicating a rather low level of COVID-19-related stress among our study participants. Moreover, the mean PSS score was around the average values (19.43 ± 5.05) (Table 1). Mean AT score was somewhat above average (43.94 ± 10.87) which implied on slightly increased level of COVID-19-related worry (Table 1).

Correlations

Later bedtime during the pandemic correlated with a higher AT and PSS scores, younger age, drinking alcohol and smoking cigarettes, higher level of education, not being partnered, not being employed, having COVID-19 contact or symptoms and getting informed about COVID-19 from multiple sources (Supplementary Table S1).

Longer sleeping time during the pandemic correlated with a higher PSS score, being male and younger, not having chronic illnesses, lower level of education, not being employed, having lower income, drinking alcohol during pandemic, not having contact with COVID-19 positive persons, not having COVID-19 personally and not getting informed about COVID-19 (Supplementary Table S1).

Good sleep quality during the pandemic correlated with a higher AT score, lower PSS and CSS scores, being younger, being male and having no chronic illnesses, not being partnered, lower education, not being employed, but not having money loss during the pandemic, avoiding smoking, drinking alcohol and immunity boosters, not having COVID-19 contact or symptoms and not getting informed about COVID-19 (Supplementary Table S1).

Regression models

Significant models with acceptable fit (moderate adjusted R²) were obtained for the Total CSS score based on the investigated sleeping patterns (Tables 5 and 6). Multicolinearity analysis showed slight to moderate correlations between variables that were not relevant for the model fit (all VIF < 3). Homoscedasticity of residuals was visually confirmed by scatter plot inspection.

Table 5.

Unadjusted linear regression model representing the association of sleeping patterns with a higher COVID Stress Scales score

Parameters	B	Standard Error	Variance inflation factor	Lower 95% CI for B	Upper 95% CI for B	p
Constant	42.418	4.143		34.294	50.541	0.001
Bedtime before pandemic	−3.327	2.054	1.280	−7.354	0.701	0.105
Bedtime in pandemic	2.488	2.135	1.271	−1.700	6.675	0.244
Sleep duration before pandemic	0.321	0.718	2.444	−1.087	1.728	0.655
Sleep duration in pandemic	1.080	0.774	2.418	−0.437	2.597	0.163
Sleep schedule in pandemic compared to pre-pandemic	2.738	0.488	1.100	1.780	3.696	0.001
Sleep quality before pandemic	0.742	0.833	1.091	−0.892	2.376	0.373
Sleep quality in pandemic	−7.266	0.791	1.224	−8.817	−5.715	0.001

Pandemic – COVID-19 pandemic; CI – confidence interval. Bolded values represent statistical significance

Table 6.

Adjusted linear regression model representing the association of sleeping patterns with a higher COVID Stress Scales score

Parameters	B	Standard Error	Variance inflation factor	Lower 95% CI for B	Upper 95% CI for B	p
Constant	111.124	21.980		67.957	154.290	0.001
Bedtime before pandemic	−3.884	3.174	1.623	−10.119	2.350	0.222
Bedtime in pandemic	3.668	3.340	1.557	−2.892	10.228	0.273
Sleep duration in pandemic	0.990	1.293	3.062	−1.549	3.529	0.444
Sleep duration before pandemic	0.451	1.166	3.056	−1.839	2.741	0.699
Sleep schedule in pandemic compared to pre-pandemic	2.273	0.676	1.193	−0.054	2.601	0.046
Sleep quality before pandemic	0.277	1.391	1.499	−2.455	3.009	0.842
Sleep quality in pandemic	−3.283	1.303	1.756	−5.841	−0.725	0.012
AT score	−1.046	0.075	1.466	−1.193	−0.899	0.001
PSS score	0.820	0.177	1.454	0.473	1.167	0.001
Gender	0.257	1.605	1.853	−2.894	3.409	0.873
Age	−2.107	0.058	1.487	−4.220	1.006	0.036
Chronic illness	2.661	1.046	1.269	0.606	4.715	0.011
Weight change in pandemic	−0.871	0.993	1.718	−2.821	1.079	0.381
Relationships	−0.407	1.704	1.288	−3.754	2.940	0.811
Education level	−1.223	1.391	1.277	−3.954	1.508	0.379
Employment status now	−3.383	2.653	1.175	−8.594	1.828	0.203
Employment before pandemic	−0.491	2.576	1.518	−5.550	4.568	0.849
Income	−1.627	0.002	1.134	−2.001	0.013	0.382
Money loss in pandemic	1.558	1.798	1.382	−1.973	5.089	0.387
Smoking cigarettes	−3.208	19.057	1.139	−8.634	−1.782	0.024
Smoking in pandemic	2.161	1.334	1.116	−0.458	4.780	0.106
Alcohol drinking	−4.408	1.905	1.136	−8.148	−0.668	0.021
Alcohol in pandemic	2.957	1.399	1.100	0.209	5.705	0.035
COVID-19 contact	−0.435	1.186	1.909	−2.764	1.894	0.714
Suspicious contact	0.356	1.127	1.489	−1.858	2.570	0.752
COVID-19 in family	4.583	2.010	1.531	0.635	8.531	0.023
Having COVID-19	−3.132	3.095	1.384	−9.211	2.947	0.312
COVID-19 symptoms	−0.478	1.932	1.700	−4.272	3.317	0.805
Immunity boosters use	3.651	0.913	1.314	1.860	5.443	0.001
Information	0.662	0.809	1.222	−0.926	2.250	0.413

Pandemic – COVID-19 pandemic; CI – confidence interval. Bolded values represent statistical significance

Table 5 showcases univariate (unadjusted) linear regression model. If only sleeping patterns were assessed (unadjusted model), a higher CSS score was associated with sleeping more often compared to pre-pandemic sleeping schedule and poor sleep quality during the pandemic (R = 0.620; adjR²=0.581; p = 0.001).

When models were adjusted for the AT and PSS scores as well as socio-demographic parameters, it can be seen that sleeping more often compared to pre-pandemic sleeping schedule and poor sleeping quality during the pandemic remained independently associated with a higher CSS (Table 6). Also, both AT and PSS scores as well as younger age, having chronic illnesses, not being a smoker before the pandemic, not drinking alcohol before the pandemic, but drinking alcohol during the pandemic, using immunity boosters and having COVID-19 in family were associated with a higher CSS score (R = 0.683; adjR²=0.642; p = 0.001) (Table 6).

Discussion

This study revealed that the majority of study participants from Serbia and Republic of Srpska (Bosnia and Herzegovina) did not report changes in sleep duration over the third and the fourth epidemic wave. Few people slept poorly before the pandemic, but the prevalence of people who reported poor sleep quality doubled during the pandemic compared to pre-pandemic period (4.5% vs. 9.4%). About one in ten people reported sleeping more often compared to pre-pandemic sleeping schedule. Overall, sleeping more often compared to pre-pandemic sleeping schedule and having poor sleep quality predicted higher COVID-related stress. These findings suggest that sleep quality is an important element in being resilient to stress in crises. Triage of people who change their sleep patterns, sleep more often and have impaired sleep quality in the pandemic might help to enhance coping with stress by introducing behavioral changes and changes to their environment [15, 16].

There is compelling evidence that the COVID-19 pandemic affected people’s sleep [17]. While the degree of sleep disturbances varied, some people experienced insomnia, daytime sleepiness, sleep apnea, had abnormal dreams or other sleep dysfunctions [18]. Many people globally changed their bedtime and went to sleep later than before the pandemic [17, 19]. Only about 12% of people in this study changed their sleep duration during the pandemic, whereas other authors found that sleep duration shortened, on average, by one hour and sleep quality worsened [20]. Consequently, people slept more frequently, mainly during daytime, to catch up on reduced sleep [20].

This study observed that a small proportion of people slept more often during the third and the fourth COVID-19 epidemic waves compared to pre-pandemic schedule. Fear of catching the virus has been observed as one of the major anxiety-inducing factors [10] that could be associated with sleeping more often compared to pre-pandemic sleeping schedule to manage stress. Furthermore, loss of sleep and poor sleep quality might be associated with government-imposed control measures, such as opening hours of cafes and restaurants until 5 pm and use of COVID-19 certificates when about 50% of the population was vaccinated with at least one vaccine dose. Activities outside the home have probably been replaced by following of news and media outlets on TV and the Internet, so people more often stayed up for longer and were exposed to screens [10, 21]. Furthermore, the infodemic, or influx of a large amount of information about the virus [10], as well as daily reports about hospitalizations and deaths likely affected sleep of some participants, as it has been observed that media reporting particularly among older people was interpreted as bothersome [22]. Similarly, stress associated with the economic losses might have perpetuated sleep dysfunction and poor sleep quality, as many people had to reduce business hours or close down their businesses to lower the risk of viral spreading [17].

Other characteristics of study participants relevant for sleep disturbances and higher COVID-related stress in the third and the fourth pandemic wave warrant further discussion. For example, younger people were more likely to have poorer sleep. This finding is in line with previous literature [23, 24]. A study among young adults in the Western Balkans, including Serbia, suggested that more than one-half of them had some sleep changes, particularly shorter duration of sleep [23]. However, unlike among younger people [23], there was no difference between men and women in the general population sample in this study. A multinational study similarly found that older people had better sleep quality in the pandemic [24]. This is being explained by self-perceived high risk of COVID-19 and economic hardship that affected younger people more often [24, 25]. But, these factors also seem to be associated with other health behaviors during the pandemic, such as eating poorer diet, and increasing tobacco smoking and alcohol intake [24].

Not smoking and not drinking alcohol before the pandemic as well as increased alcohol drinking during the pandemic among people who drank alcohol before the pandemic was also associated with COVID-19-related stress. Increased alcohol drinking during the pandemic has been observed in previous studies [24, 26]. Evidence suggests that that people who experienced symptoms of anxiety and depression were more likely to drink alcohol during the pandemic compared to people without either of those symptoms [26]. This finding was more pronounced among adults younger than 40 years of age [26], highlighting maladjustment and unhealthy coping with the everyday dynamics during the pandemic. In a multi-national study across Europe and the US, it has been found that more than 80% did not change their drinking behavior during the pandemic. In fact, one in 8 people reported drinking less and 5% reported drinking more compared to pre-pandemic [27]. In this study, we observed that, compared to pre-pandemic period, practically one in three people began drinking less during the pandemic, while almost one in 10 drank more. Although this category of people, who drank more during the pandemic, was the least prevalent one, it raises an important issue of coping mechanisms when under stress that should be acknowledged in future health crises.

Finally, having chronic illnesses and family with COVID-19 was associated with a higher stress level due to COVID-19, which was expected. As people with pre-existing chronic illnesses were one of the population groups at risk of severe COVID-19 forms and poorer COVID-19 outcomes [28], they may have been more adherent to the recommended prevention measures, such as wearing masks and avoiding social events and indoor and outdoor gatherings, to reduce exposure for fear of catching the virus and potentially having severe or fatal COVID-19 [29]. Evidence suggests that poorer health status was directly affecting sleep quality and sleep duration [30]. Also, people whose family members were affected by COVID-19, especially as informal caregivers within the nuclear family experienced higher risk of anxiety associated with the anticipated infection outcomes of the loved ones [29, 30], but also own fear of catching COVID-19. All factors mentioned above corroborate the proposed framework of predisposing (health status, pre-existing chronic illnesses), precipitating (fear for family and loves ones, economic strains) and perpetuating factors (anxiety about the pandemic) that have detrimental effect on sleep during the pandemic [19].

It should also be mentioned that a phenomenon observed during the pandemic – social jetlag i.e. delayed bedtime during weekdays relative to days off from work – could have had a role in experience of COVID-19-related stress through disruption of sleep-wake cycle. Delayed bedtime has been associated with insomnia [31]. Namely, because of social distancing and reduced social interactions due to shorter opening times of public spaces (such as cafés and restaurants), people were staying at home and were more likely to go to sleep later in the night. This was especially pronounced among the unemployed people and people who lost their jobs during the pandemic and people who were working from their homes [32, 33]. It has been observed that working longer hours was associated with social jetlag [34], which could be explained by missing social interactions. Bearing in mind that in our study one quarter of people sustained money loss and one third were unemployed, these events could also perpetuate later bedtime, insomnia and daytime napping associated with stress of having to sustain their livelihood.

Study limitations

Some limitations of this study should be kept in mind when interpreting the results. Data collection coincided with the third (November 2020-March 2021) and the fourth epidemic wave (July-September 2021). As a result, data were collected in the period of lower occurrence of COVID-29 in between epidemic waves. This could have influenced the perception of sleeping patterns rendering them heterogeneous throughout the period of data collection. All data in this study were self-reported, so information bias may be an issue because data collection spanned over one year, so it may be difficult to assess precisely sleeping patterns before the pandemic. Surveys such as this one often have to be adjusted for time constraints when people in the general population are asked to fill in the questionnaires. As a result, the questionnaire had to be reduced to the most relevant data. In line with this, no specific validated sleep quality questionnaire was used to assess sleep quality, as ordinal responses on sleep quality were measured by one single questionnaire item. That is why some pieces of information may have been omitted. With regards to participant selection, as some people were able to work through administration affairs online, we were unable to capture the entire population that required municipality services at the time of survey. Therefore, our results cannot be generalized to the entire population of Serbia and Republic of Srpska. This is a cross-sectional study, where data on exposures and outcomes are collected at the same time, so temporal sequence cannot be established and causal association may not be true.

Public health implications

Bearing in mind that one in four participants had changes in sleeping schedule during pandemic compared to the pre-pandemic period and that there was an increase in poor sleep quality, population-level implications for monitoring sleep during public health crises have become increasingly relevant. Changes in sleep patterns and poor sleep quality are closely associated with stress and anxiety, which has repercussions on public mental health, work-related productivity and well-being of individuals, families and communities. In future health crises, it is of paramount importance to raise awareness through traditional (TV, radio, newspapers) and contemporary media (Internet, YouTube, Instagram) about sleep health and recommendations on how to promote sleep health as best as possible. Specific population groups prone to poor sleep quality such as the unemployed, people working in shifts, and those who are largely socially isolated (e.g. older people or people with disabilities) should be prioritized in efforts to foster public mental health in crises such as the pandemic.

In conclusion, about one-quarter of participants from Serbia and Republic of Srpska – Bosnia and Herzegovina in this study had changes in sleep patterns such as shorter or longer sleep duration or were sleeping more often compared to pre-pandemic sleeping schedule. Overall, bedtime was not changed. Proportion of participants who labeled their sleep quality as poor doubled in the pandemic – with about one in 10 people reporting poor or very poor sleep quality. Sleeping more often compared to pre-pandemic sleeping schedule and poor sleep quality were independent factors associated with a higher level of COVID-19-related stress throughout the second pandemic year. Optimizing sleep quality in crises, such as the pandemic, in people who experience poor sleep quality, should be prioritized when managing public health emergencies at a national and international level.

Author contributions

BJ contributed to study design, data collection, analysis and interpretation and drafted the manuscript. JD, MM, MK, DB, SR, JS, ZSR, DL, JF, VN, MD, MC, DNŽ and TG contributed to study design, data collection, analysis and interpretation and provided critical review of the manuscript. All authors approved the final version of the manuscript before submission and agree to be held accountable for all aspects of this research.

Funding

Article processing charges or other similar costs applied by the publisher to provide for open access publication will be met by the Faculty of Medicine – Foca, University of East Sarajevo.

Data availability

The dataset underlying this study is available upon request to the corresponding author.

Declarations

Ethics approval and consent to participate

The Ethics Committee of the University of Pristina temporarily settled in Kosovska Mitrovica approved the study (No. 2179). All participants provided written informed consent for the study. The study was conducted in line with the principles outlined in the Declaration of Helsinki. All participants signed informed consent prior to enrolment.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.