Prevalence and factors associated with severe fatigue 2 years into the COVID-19 pandemic: a cross-sectional population-based study in Geneva, Switzerland

Abstract

Objectives

This study aims (1) to assess the prevalence of severe fatigue among the general population of Geneva, 2 years into the COVID-19 pandemic and (2) to identify pandemic and non-pandemic factors associated with severe fatigue.

Design

Cross-sectional population-based survey conducted in Spring 2022.

Setting

General adult population of Geneva, Switzerland.

Participants

6870 adult participants, randomly selected from the general population, included in the Specchio-COVID-19 cohort study, were invited to answer an online health survey.

Outcome and cofactor measure

Prevalence of severe fatigue was measured by the Chalder Fatigue Questionnaire with a cut-off score≥4 out of 11. We assessed prevalence ratios of severe fatigue considering sociodemographic factors, health and behavioural characteristics (body mass index, depression, recent diagnosis of chronic disease or allergy, acute health event, smoking status, physical activity and sleep quality) and recent self-reported COVID-19 infections.

Results

A total of 4040 individuals participated (participation rate 59%, 58% were women, mean age 53.2 (SD=14.1 years)). Overall prevalence of severe fatigue was 30.7% (95% CI=29.2%–32.1%). After adjusting for age, sex, educational level and pre-existing comorbidities, the following characteristics were associated with severe fatigue: individuals aged 18–24 years (adjusted prevalence ratio (aPR)=1.39 (1.10–1.76)) and 25–34 years (aPR=1.23 (1.05–1.45)), female sex (aPR=1.28 (1.16–1.41)), depression (aPR=2.78 (2.56–3.01)), occurrence of health events unrelated to COVID-19 (aPR=1.51 (1.38–1.65)) and self-reported COVID-19 infection in the past 12 months (aPR=1.41 (1.28–1.56)). After further adjustment for depression, previous associations were maintained except for young age.

Conclusions

About one-third of the adult general population of Geneva experienced severe fatigue, 2 years into the COVID-19 pandemic. Heightened fatigue among young adults is partly explained by depressive symptoms. Recent COVID-19 infection is substantially associated with severe fatigue, regardless of infection severity or co-occurrence of depressive disorder.

Trial registration number

CCER project ID 2020-00881.

Strengths and limitations of this study.

• Fatigue measurement with a validated scale in a large randomly selected sample from the general population.

• Application of weighting to mitigate non-response bias and selection bias.

• Under-representation of vulnerable groups (low incomes, limited education and non-Swiss residents) and older frail adults.

• Cross-sectional design prevented us from making causal inferences.

Introduction

Fatigue is a common symptom with a negative impact on quality of life at individual level, but also on healthcare system and society. Fatigue is defined as a feeling of exhaustion (physical, cognitive or emotional) or lack of energy affecting the performance of usual daily activities, and not relieved by resting.^{1 2} Fatigue can be classified by duration as recent (within the last month), prolonged (lasting 1–6 months) or chronic (lasting more than 6 months).¹ Despite its subjectivity, various scales measure fatigue severity, such as the Chalder Fatigue Questionnaire (CFQ) which qualifies fatigue as severe with a score of 4 or more out of 11 items.

Individuals reporting severe fatigue experience a range of negative outcomes, including lower employment rates,³ reduced quality of life,⁴ elevated risks of cardiovascular disease,^{5 6} cancers,⁷ depression^{8 9} and increased mortality.⁷ Moreover, fatigue contributes to heightened primary-care consultations and imposes a disproportionate burden of disease.¹⁰ One-third of fatigue cases lack a clear aetiology,¹ although chronic conditions and psychosocial issues such as depression or anxiety are often identified in primary-care setting .^{11 12}

The COVID-19 pandemic had major economic and societal consequences fostering sedentary lifestyle and increased loneliness due to isolation policies.¹³ These factors collectively impacted mental health,¹⁴ potentially raising severe fatigue levels in Geneva. Furthermore, since the beginning of the pandemic in February 2020, over 200 000 infections have occurred in the canton of Geneva,¹⁵ and a previous serosurvey showed that 72% of the population had been infected by SARS-CoV-2.¹⁶ Although the majority of non-hospitalised people fully recover, 20%–46% report persistent fatigue months later.17,19 Fatigue is also a frequent symptom of postacute COVID-19 syndrome, defined as signs or symptoms that develop subsequently to SARS-CoV-2 infection and persist for at least 12 weeks.²⁰ The cumulative incidence of postacute COVID-19 syndrome ranged from 3.5% to 10.5% per 100 persons depending on SARS-CoV-2 serotype and vaccination status.²¹ As SARS-CoV-2 remains actively circulating within the population, COVID-19 infections may have modified the prevalence of severe fatigue and would impact the populational epidemiological landscape for years to come.

Previous studies examined fatigue prevalence in the general population before the COVID-19 pandemic.^{22 23} There has been limited exploration of fatigue prevalence and its associated factors, including those specific to COVID-19, in the general population, especially during the pandemic. We hypothesised that fatigue prevalence would be elevated in the general population, given the direct and indirect effects of the COVID-19 pandemic.

This study aims (1) to assess the prevalence of severe fatigue among the general adult population of Geneva, Switzerland, 2 years into the COVID-19 pandemic and (2) to identify pandemic and non-pandemic specific factors associated with severe fatigue.

Methods

Study setting

This cross-sectional study was conducted as part of Specchio-COVID-19, a population-based digital study launched in December 2020 to follow-up serosurvey participants in Geneva, Switzerland.²⁴ The study methods, including study design and detailed participant selection, have been described in previous publication.²⁴

Study population

Serosurvey participants were randomly selected from Bus Santé population-based study,25,27 and from Geneva state registries between April 2020 and July 2021.^{16 28 29} Serosurveys occurred in three periods of several weeks within this timeframe, and each participant had at least one measure of anti-SARS-CoV-2 antibodies at inclusion.¹⁶ Exclusion criteria were limited to individuals residing in institutions, such as nursing homes, and those who were imprisoned.²⁸ After participating in at least one serosurvey, all adult participants were invited to take part in Specchio-COVID-19 longitudinal study and completed an inclusion questionnaire online.²⁴ Between March and May 2022, an annual general health follow-up questionnaire, which included fatigue assessment, was proposed online to all participants of the Specchio-COVID-19 cohort (n=6870 individuals). To reduce the amount of non-response, a reminder to answer was sent 2 weeks after the questionnaire was sent out. Participants were also given the option of completing a paper version.

Participation rate of the annual general health follow-up questionnaire was 58.8% (4040/6870). Respondent characteristics compared with general population and non-respondent characteristics are available in onlinesupplemental tables S1 S2. Compared with non-respondents, young and middle-aged adults (18–49 years old) and people with primary education were under-represented among respondents. Compared with the Geneva general population, there was an over-representation of people in the 45–64 age group (49.3% vs 34.4%), while the proportion of young adults was lower in our sample (24.5% vs 44.6%). Swiss citizens (84.2% vs 63.6%), and people with tertiary education (65.0% vs 40.2%), were also over-represented in the study population.

Data collected

Data on sociodemographic characteristics (age, sex, education level, household income, professional status and household composition), the presence of chronic diseases and smoking status were collected at inclusion. Physical activity and sleep quality were assessed in another dedicated questionnaire (behavioural questionnaire) launched in May 2022. Other variables were collected in the annual general health follow-up questionnaire. All data were self-reported.

Measures

Fatigue assessment

The main outcome was the presence or absence of severe fatigue measured by the CFQ-11.³⁰ The CFQ-11 is an 11-item self-reported questionnaire describing physical fatigue (lack of energy, feeling weak, reduced muscle strength and need to rest), and mental fatigue (concentration, cognitive performances and memory) in the last 4 weeks. It has been validated in hospital and community populations.30,32 Binary scoring was applied to discriminate between presence or absence of severe fatigue with a 75.5% sensitivity and 74.5% specificity (cut-off≥4 out of 11).³⁰ Internal consistency of the CFQ was high in our study (Cronbach’s alpha=0.92).

Factor analyses

Factors were chosen after literature review on fatigue predictors. These factors were categorised as sociodemographic factors, and health and behavioural factors.

Sociodemographic factors

Sociodemographic variables included age, sex, employment status (employed, self-employed, unemployed, retired or other economically inactive (students or people without any professional activity)) and educational level (primary, secondary or tertiary). Household income was categorised according to household composition, using the latest information from the Cantonal Office of Statistics of Geneva for 2015–2017 (low (below the first quartile of the income distribution), medium (between the first and third quartiles) or high (above the third quartile)). Age groups (18–24, 25–34, 35–44, 45–54, 55–64, 65–74 and 75–98 years old) were chosen to allow comparison with previous studies.

Health and behavioural factors

Health-related factors included chronic disease diagnosis within the past year (binary), body mass index (BMI) (below 18.5, 18.5–24.9, 25.0–29.9, 30.0 and above), health event not related to COVID-19 within the past year (hospital stay, accident requiring medical assistance and pregnancy). Depression was detected using the Patient Health Questionnaire-2 (PHQ-2),³³ which assesses the frequency of depressed mood and anhedonia over the past 2 weeks. A threshold of 3 or above provides a 61% sensitivity and a 92% specificity for the diagnosis of major depression.³³ Cronbach’s alpha was 0.79 for the PHQ-2 scale in our study.

Behavioural factors included smoking status (non-smoker, former smoker and current smoker), sleep quality (dichotomised as good to very good or poor to very poor) and physical activity (active, partially active and sedentary). Based on physical activity guidelines (Federal office of sports in Switzerland, OFSPO),³⁴ participants who engaged at least five times per week in 30 min of moderate activity (defined as activity requiring moderate effort that significantly accelerates heart rate) or at least twice per week in vigorous activity (defined as activity that induces sweating) were classified as being active. Participants who engaged between once to four times per week in moderate activity or once a week in vigorous activity were classified being as partially active. Participants who did not meet either of these criteria were classified as being sedentary.

Self-reported COVID-19 infections

Participants were asked whether they had tested positive for SARS-CoV-2 in the last 12 months (via PCR test, antigenic test or self-antigenic test), regardless of symptom occurrence. The definition of SARS-CoV-2 infection relied on participant self-reporting, irrespective of the presence of antibodies at the time of serosurvey inclusion, which dated back more than 2 years for some participants. This approach was based on several assumptions. First, we assumed that most participants had infection-induced antibodies at the time of study, as indicated by seroprevalence data from 2022.¹⁶ Second, the risk of new onset postacute COVID-19 syndrome decreases as time since infection increases.³⁵ Given that fatigue is a major symptom of postacute COVID-19 syndrome, we aimed to accurately capture recent infections while avoiding misclassification of individuals with earlier infections who had fully recovered. We also conducted subgroup analyses to examine the association between severe fatigue and recent COVID-19 infection, excluding individuals who self-reported postacute COVID-19 syndrome.

Statistical analyses

All respondents (n=4040) were included in the statistical analyses. Figure 1 shows the flowchart for participation and management of missing data. Among participants (n=4040), 728 (18%) participants did not respond to their income category at inclusion. Other missing data were found in the following categories: physical activity (n=782, 19%), sleep quality (n=782, 19%), education level (n=6), BMI (n=2) and working situation (n=1). Multiple imputation was therefore performed for these variables assumed to be missing at random or completely at random, using the R package mice, V.3.16.0. Fifty mice imputed data sets were created. Variables used for the imputation procedure were age, sex, income, education level, working situation, household composition, presence of children under 5 years of age, pre-existing comorbidities, smoking status and BMI. A significant number of eligible participants (n=2830) did not answer the annual general health questionnaire including fatigue assessment. To reduce the effect of a potential non-response bias, we then used inverse probability weighting for each of the 50 imputed data sets to weight the data collected from those who did reply according to the characteristics of those who did not reply. Variables used for the weighting included age, sex, income, education level, working situation, household composition, presence of children under 5 years of age, pre-existing comorbidities and BMI. The models (see below) were run on each of the 50 multiply imputed and weighted data sets, with results combined using Rubin’s rules.

Figure 1. Flowchart showing participation and management of missing data.

Associations between severe fatigue and each factor were evaluated in an unadjusted model using Pearson’s Chi² test (table 1). Then, we performed quasi-Poisson regression with log link function to assess associations in two different adjusted multivariable models. Model 1 included an adjustment with age, sex, education level and pre-existing comorbidities, while model 2 was additionally adjusted for depression. Results of models 1 and 2 were expressed in adjusted prevalence ratios (aPR) with 95% CI. Two-tailed significance level of α = 0.05 was used for statistical testing. We also tested the interaction effect between sex and each factor on severe fatigue.

Table 1. Characteristics of study participants, overall and by fatigue status.

Sociodemographic characteristics	Non-missing data (n)	Overall, n (%)		Not feeling severe fatigue, n (%)		Feeling severe fatigue, n (%)		P value*
Age (years), mean (SD)	4040	53.2	(14.1)	54.7	(13.9)	50.0	(14.0)	<0.001
18–24		95	(2.3)	51	(1.8)	44	(3.6)	<0.001
25–34		271	(6.7)	158	(5.6)	113	(9.1)
35–44		623	(15.4)	387	(13.8)	236	(19.0)
45–54		1005	(24.9)	663	(23.7)	342	(27.6)
55–64		987	(24.4)	699	(25.0)	288	(23.2)
65–74		745	(18.4)	597	(21.3)	148	(11.9)
75–98		314	(7.8)	246	(8.8)	68	(5.5)
Sex	4040							<0.001
Female		2333	(57.9)	1529	(54.6)	804	(64.9)
Male		1696	(42.0)	1266	(45.2)	430	(34.7)
Other		11	(0.3)	6	(0.2)	5	(0.4)
Education	4034							0.8
Primary		146	(3.6)	104	(3.7)	42	(3.4)
Secondary		1266	(31.4)	885	(31.6)	381	(30.8)
Tertiary		2622	(65.0)	1807	(64.5)	815	(65.8)
Income	3312							0.15
Low		539	(16.3)	353	(12.6)	186	(15.0)
Medium		2187	(66.0)	1542	(55.1)	645	(52.1)
High		586	(17.7)	407	(14.5)	179	(14.5)
Swiss nationality	4040							<0.001
Yes		3402	(84.2)	2403	(85.8)	999	(80.6)
No		638	(15.8)	398	(14.2)	240	(19.4)
Working status	4039							0.2
Employed		2222	(55.0)	1463	(52.3)	759	(61.3)
Retired		1025	(25.4)	817	(29.2)	208	(16.8)
Economically inactive		371	(9.2)	226	(8.1)	145	(11.7)
Self-employed		310	(7.6)	225	(8.0)	85	(6.9)
Unemployed		111	(2.7)	69	(2.5)	42	(3.4)
Household composition	4040							<0.001
Couple without children		1256	(31.1)	954	(34.1)	302	(24.4)
Couple with children		1678	(41.5)	1125	(40.2)	553	(44.6)
Living alone		618	(15.3)	424	(15.1)	194	(15.7)
Living with other people		261	(6.4)	161	(5.7)	100	(8.1)
Single parent with children		227	(5.6)	137	(4.9)	90	(7.3)
Health and behavioural characteristics
BMI (kg/m2)	4038							0.030
Underweight (BMI<18.0)		127	(3.1)	93	(3.3)	34	(2.7)
Normal weight (BMI=18.0–24.9)		2437	(60.4)	1696	(60.6)	741	(59.9)
Overweight (BMI=25.0–29.9)		1120	(27.7)	789	(28.2)	331	(26.7)
Obesity (BMI≥30.0)		354	(8.8)	222	(7.9)	132	(10.7)
Health event not related to COVID-19 over the past 12 months	4040							<0.001
No		2668	(66.0)	1994	(71.2)	674	(54.4)
Yes		1372	(34.0)	807	(28.8)	565	(45.6)
Depression	4040							<0.001
No depressive disorder, PHQ-2<3		3739	(92.5)	2748	(98.1)	991	(80.0)
Major depressive disorder, PHQ-2≥3		301	(7.5)	53	(1.9)	248	(20.0)
Physical activity	3258							<0.001
Active		1750	(53.7)	1291	(56.4)	459	(47.5)
Partially active		1089	(33.4)	735	(32.1)	354	(36.6)
Sedentary		419	(12.9)	265	(11.6)	154	(15.9)
Smoking status	4040							0.058
Non-smoker		2190	(54.2)	1548	(55.3)	642	(51.8)
Former smoker		1266	(31.3)	869	(31.0)	397	(32.0)
Current smoker		584	(14.4)	384	(13.7)	200	(16.1)
Sleep quality	3258							<0.001
Good or very good		2480	(76.1)	1875	(81.8)	605	(62.6)
Bad or very bad		778	(23.9)	416	(18.2)	362	(37.4)
Self-reported COVID-19 infection	4040							<0.001
No		3266	(80.8)	2363	(84.4)	903	(72.9)
Yes		774	(19.2)	438	(15.6)	336	(27.1)

^*Pearson’s Chi-squaredχ2 test.

BMIbody mass index

All data analysis and modelling were performed using R V.4.3.1.³⁶ The survey package version 4.2.1 was used for incorporating non-response weights into the prevalence ratio model.³⁷

We used the STrengthening the Reporting of Observational Studies in Epidemiology (STROBE) cross-sectional reporting guidelines to ensure the quality of this study.³⁸

Participants and public involvement

Participants and public were not involved in study design or analyses. Nevertheless, the Specchio-Hub platform (formerly Specchio-COVID-19 platform, www.specchio-hub.ch) provided up-to-date scientific content in plain language on the COVID-19 pandemic and its impact on Geneva population. Participants were also invited to give their opinions on received surveys and had easy access to scientific experts by e-mail, telephone or by answering queries online. Participants received regular newsletters on research process and results. They could also take part in free open-access webinars with study’s principal investigators.

Results

A total of 4040 participants participated in the survey (participation rate 59%). Table 1 shows respondent characteristics stratified by fatigue status. Among respondents, 58% were women with a mean age of 53.2 years (SD=14.1 years). Most of the respondents were Swiss nationals (84%), completed a tertiary education (65%) and were employed (55%) or retired (25%). A large proportion of them were non-smokers (54%) and had an active lifestyle (54%), whereas 19% declared a COVID-19 infection in the past 12 months.

Unadjusted analyses showed that young adults (25–34 years old), women, economically inactive people, people with low income, single parents with children, couples with children, people living alone or with others but not a partner and parents having children under 5 years of age, had greater prevalence of severe fatigue. Similarly, health and behavioural characteristics associated with severe fatigue included obesity, inactive or partially active lifestyles, poor sleep quality, occurrence of health events not related to COVID-19 within the past year, new chronic disease diagnosis within the past year and depression. People reporting COVID-19 infection in the past year also had a higher risk of severe fatigue.

Overall prevalence of severe fatigue was 30.7% (95% CI=29.2%–32.1%). The highest proportion of people experiencing severe fatigue was observed in the 18–44 age group (39.7%, 95% CI=38.2%–41.2%), while the lowest was in the 65–98 age group (20.4%, 95% CI=19.1%–21.6%).

Results of the adjusted regression models (models 1 and 2) are shown in table 2. In the multivariable weighted model adjusted for age, sex, education and chronic diseases (model 1), severe fatigue was strongly associated with depression (aPR=2.78 (2.56–3.01)), poor sleep quality (aPR=1.58 (1.43–1.74)) and the occurrence of health events not related to COVID-19 (aPR=1.51 (1.38–1.65)). Moderate associations were seen between severe fatigue and young age (18–34 years), female sex, single parent with children, obesity, inactive behaviour, recent diagnosis of a chronic disease and self-reported COVID-19 infection in the past 12 months (aPR ranged from 1.25 to 1.41 with 95% CI=1.05–1.76). Significant associations with severe fatigue, though to a lesser extent, were also observed among individuals living alone, former and current smokers, as well as individuals demonstrating partially active behaviour (aPR ranged from 1.15 to 1.17 with 95% CI=1.00–1.35). Middle and old age were inversely associated with severe fatigue (aPR range from 0.56 to 0.85 with 95% CI=0.48–0.97). No association was found between severe fatigue and income categories, education levels or working status.

Table 2. Quasi-poisson multivariable regression models of each factor on severe fatigue.

	Model 1:adjusted aPR with MI (95% CI) and IPW*	Model 2:adjusted aPR with MI (95% CI) and IPW†
Sociodemographic characteristics
Age
45–54	Ref.	Ref.
18–24	1.39 (1.10–1.76)‡	1.21 (0.97–1.52)
25–34	1.23 (1.05–1.45)‡	1.16 (0.99–1.35)
35–44	1.13 (0.99–1.29)	1.12 (0.99–1.28)
55–64	0.85 (0.75–0.97)‡	0.88 (0.78–1.00)
65–74	0.56 (0.48–0.67)‡	0.61 (0.52–0.72)‡
75–98	0.61 (0.48–0.77)‡	0.66 (0.53–0.82)‡
Sex
Male	Ref.	Ref.
Female	1.28 (1.16–1.41)‡	1.24 (1.13–1.36)‡
Education
Tertiary	Ref.	Ref.
Primary	0.92 (0.71–1.20)	0.84 (0.65–1.08)
Secondary	0.99 (0.89–1.09)	0.96 (0.87–1.06)
Income
High	Ref.	Ref.
Low	1.09 (0.92–1.29)	1.06 (0.90–1.25)
Medium	0.99 (0.86–1.14)	0.99 (0.86–1.13)
Working status
Employed	Ref.	Ref.
Economically inactive	0.98 (0.84–1.15)	0.89 (0.76–1.03)
Unemployed	1.09 (0.86–1.39)	0.99 (0.79–1.26)
Self-employed	0.88 (0.73–1.07)	0.84 (0.70–1.01)
Retired	0.76 (0.58–1.00)	0.72 (0.55–0.95) ‡
Household composition
Couple without children	Ref.	Ref.
Single parent with children	1.30 (1.06–1.59)‡	1.23 (1.02–1.49)‡
Couple with children	1.09 (0.95–1.25)	1.14 (1.00–1.29)
Living alone	1.16 (1.00–1.35)‡	1.14 (0.99–1.32)
Living with other people	1.12 (0.89–1.40)	1.02 (0.82–1.26)
Having any young children below 5
No	Ref.	Ref.
Yes	1.11 (0.94–1.30)	1.15 (0.98–1.35)
Health and behavioural characteristics
BMI
Normal weight (BMI=18.0–24.9 kg/m2)	Ref.	Ref.
Underweight (BMI<18.0)	0.81 (0.60–1.08)	0.83 (0.63–1.09)
Overweight (BMI=25.0–29.9)	1.06 (0.95–1.18)	1.07 (0.96–1.18)
Obesity (BMI≥30.0)	1.25 (1.08–1.44)‡	1.18 (1.02–1.36)‡
Chronic disease or allergy diagnosis within the past year
No	Ref.	Ref.
Yes	1.42 (1.29–1.56)‡	1.32 (1.21–1.45) ‡
Health event not related to COVID-19 within the past year§
No	Ref.	Ref.
Yes	1.51 (1.38–1.65)‡	1.44 (1.32–1.58)‡
Depression
No depressive disorder, PHQ-2<3	Ref.	Ref.
Major depressive disorder, PHQ-2≥3	2.78 (2.56–3.01)‡	2.78 (2.56–3.01)‡
Physical activity
Active	Ref.	Ref.
Partially active	1.16 (1.04–1.30)‡	1.16 (1.04–1.28)‡
Inactive	1.22 (1.06–1.42)‡	1.14 (0.99–1.32)
Smoking status
Non-smoker	Ref.	Ref.
Former smoker	1.17 (1.05–1.30)‡	1.13 (1.02–1.25)‡
Current smoker	1.15 (1.01–1.30)‡	1.07 (0.95–1.21)
Sleep quality
Good or very good	Ref.	Ref.
Bad or very bad	1.58 (1.43–1.74)‡	1.43 (1.30–1.58)‡
Self-reported COVID-19 infection
No	Ref.	Ref.
Yes	1.41 (1.28–1.56)‡	1.41 (1.29–1.56)‡

^*Adjusted prevalence ratios with multiple imputation and inverse probability weighting. Adjustments include sex, age, education level, and chronic disease.

^†Adjusted prevalence ratios with multiple imputation and inverse probability weighting. Adjustments include sex, age, education level, chronic disease, and depression.

^‡Associations statistically significant.

^§Health event not related to COVID-19 includes hospital stay, accident requiring medical assistance, and pregnancy.

aPRadjusted prevalence ratiosBMIbody mass indexIPWinverse probability weightingMImultiple imputationPHQ-2Patient Health Questionnaire-2

After further adjustment for depression (model 2), the previous associations were maintained except for young and middle age (18–34 years and 55–64 years), people living alone, being inactive or currently smoking. Retirement reached significance level with a negative association with severe fatigue. Results of model 2 are displayed in figures2 3.

Figure 2. Quasi-poisson multivariable regression for sociodemographic factors (model 2) adjusted for sex, age, education level, chronic disease and depression. aPR, adjusted prevalence ratios.

Figure 3. Quasi-poisson multivariable regression for health and behavioural factors (model 2) adjusted for sex, age, education level, chronic disease and depression. *Chronic disease or allergy newly diagnosed within the past year. **Health event not related to COVID-19 within the past year. ***Self-reported COVID-19 infections. aPR, adjusted prevalence ratios; BMI, body mass index.

Subgroup analyses excluding individuals who self-reported postacute COVID-19 syndrome yielded similar results concerning the association between severe fatigue and recent COVID-19 infection.

Further age-adjusted and sex-adjusted standardisation of the study sample to the general population in Geneva showed similar results, including overall prevalence of severe fatigue and associations between severe fatigue and each analysed factor.

No interaction was found between sex and any of analysed factors concerning severe fatigue.

Discussion

Key results

This study showed that about one-third of the adult general population of Geneva experienced severe fatigue, 2 years into the COVID-19 pandemic. Main factors associated with severe fatigue were depression, poor sleep quality, recent health events not related to COVID-19, female sex, being a single parent with children, obesity and inactive behaviour. While these associations are in line with previous research,923 39,43 this study provides updated prevalence of fatigue in the general population after the COVID-19 pandemic and an explanation into some of the pathways leading to severe fatigue.

Interpretation

Prevalence of severe fatigue

Overall prevalence of severe fatigue was 31% in our study, which appears similar or higher compared with previous prepandemic studies. Pawlikowska et al⁴² reported a prevalence of severe fatigue of 38% in young and middle-aged adults (18–45) measured with the CFQ-11, similar to the 40% prevalence, we observed in the same age group. Similarly, a Norwegian fatigue survey⁴⁰ conducted in the general population found a severe fatigue prevalence of 22% using the same scale, which is substantially lower than the prevalence we found. A meta-analysis,²² which involved 288 000 participants worldwide (with 96% assessed before the COVID-19 pandemic), revealed a significantly lower prevalence of severe fatigue. Within the European subgroup (28% of whom assessed with CFQ-11), the prevalence was found to be 12.7%. In 2006–2010, Guessous et al⁴⁴ found a prevalence of self-reported exhaustion of 5.4% (measured as part of the frailty scale) among a representative sample of adults aged over 65 in Geneva, while we found a fatigue prevalence of 20.4% in the same age group. The cross-sectional survey done by Galland-Decker et al³⁹ also provided information on prepandemic fatigue prevalence from 2014 to 2017 in middle and older-age (45–86 years) adults in the nearby city of Lausanne, Switzerland. Fatigue prevalence was 21.9% (95% CI=20.4–23.4) using the Fatigue Severity Scale (FSS), compared with 27.7% measured with CFQ-11 (95% CI=26.1–29.3) in the same age group in our study. Nevertheless, the use of a different scale precludes reliable comparison between these prevalence studies, which constitutes a strong limitation.

Factors associated with severe fatigue

Age

Results showed a higher prevalence of fatigue in younger individuals, which has also been reported in other studies.^{23 39} Our results suggest that this association is partly explained by greater depression level in young adults, as it became statistically non-significant after adjustment for depression. Mental health status is associated with various somatic symptoms,⁴⁵ including general symptoms like fatigue.⁸ It is even more important now to carefully consider the mental health status of young individuals, with studies showing that they were particularly susceptible to mental health deterioration during the early stages of the pandemic.⁴⁶

Sex

Women were more likely to experience severe fatigue in our study. These results were diverse in the literature, with some studies showing more fatigue in women,4047,49 and others not.^{39 50} A prior Specchio-COVID-19 study showed higher psychological distress in women,¹⁴ which is also related to fatigue.⁵¹ Adjustment for depression, in this current study, did not neutralise the association between fatigue and female sex. Mental health cannot by itself explain fatigue in women, which could be an expression of health, economic and gender disparities.²³

Depression

Our study confirmed the strong association between fatigue and depression found by previous research. Earlier studies attributed this to overlapping symptoms like insomnia, hypersomnia and poor concentration.⁸ Even excluding overlapping symptoms, individuals with fatigue still had higher depression prevalence, partly due to genetic and environmental factors.⁸ Other studies suggest shared biological mechanisms such as inflammation, although the precise pathway remains unclear.⁹

The COVID-19 pandemic was a substantial stress, increasing depression, anxiety and psychological distress.^{13 14} Social isolation reinforced loneliness, particularly among people living in institutions, worsening mental health.⁵² Social media use also led to fear of contagion and excessive concern about the pandemic.^{53 54} High levels of anxiety and depression were also observed in vulnerable groups due to socioeconomic insecurity and more precarious employment.⁵⁵ These factors collectively contributed to the prevalence of fatigue observed in this study.

Sedentary lifestyle and obesity

Compared with active lifestyles, sedentary and partially active lifestyles were associated with severe fatigue. The link with sedentary lifestyle, however, became non-significant after adjustment for depression, indicating that depression may mediate this association. Importantly, sedentary behaviour is an independent risk factor for increased morbidity and mortality, which cannot be fully attributed to mental health alone.⁵⁶

Regarding the role of the pandemic in this association, lockdown policies in Geneva were less restrictive than in other European countries. People were allowed outside as often as they wanted, making it unlikely that isolation measures alone increased sedentary behaviour. Previous publication showed that 65% of our population exercised at least once weekly, with only 10% reporting reduced exercise time.⁵⁷

Results also showed that obesity was associated with severe fatigue. This could be explained by both physiological and behavioural pathways⁵⁸ as observed in other studies.^{41 59 60} A dysregulation of glucose, neuroendocrine hormones and metabolic hormones, due to chronic sleep deprivation, can result in impaired glucose tolerance and weight gain.⁵⁸ Chronic sleep deprivation also promotes unhealthy eating, inactivity and further weight gain, creating a bidirectional relationship between obesity and fatigue. Obesity enhances chronic sleep restriction through complications like mental health disorders and chronic conditions.⁵⁸ Both fatigue and obesity have been linked to inflammation,^{61 62} further contributing to fatigue.

COVID-19 infections

Recent COVID-19 infections were associated with severe fatigue with a comparable magnitude to other acute health events, such as hospital stays, accidents requiring medical assistance or pregnancy. This suggests that SARS-CoV-2 infection and postacute COVID-19 syndrome can be as debilitating as these other major health events. It is now well established that COVID-19 infections can evolve into a postacute COVID-19 syndrome which comprises persistent fatigue, respiratory and neurocognitive symptoms, fatigue being the most frequently reported symptom. In Geneva, the CoviCare study¹⁷ found that 20.7% of people who had tested positive for SARS-CoV-2 reported persistent fatigue more than 7 months after the diagnosis in outpatient settings. We may hypothesise that fatigue associated with recent COVID-19 infections could be partly explained by the occurrence of postacute COVID-19 syndrome.

In fact, a continuum exists in the expression of post-COVID symptoms, ranging from postexertional malaise to chronic fatigue syndrome,¹⁹ with an escalating impact on functional capacity. In our study, we analysed individuals who self-reported recent positive SARS-CoV-2 tests, in a population that was already mostly seropositive. We found that severe fatigue was consistently linked to recent COVID-19 infection, irrespective of infection severity, asymptomatic cases or the occurrence of postacute COVID-19 syndrome. The direct effect of SARS-CoV-2 infection may be a major factor contributing to the increased prevalence of fatigue in the general population during the COVID-19 pandemic. Studies have now shown that COVID-19 infection can raise the incidence of diabetes and other chronic conditions.^{63 64} Additional contributors to fatigue include depression, with results in this study confirming the association between depression and fatigue. The presence of depression however did not weaken the association between SARS-CoV-2 infection and fatigue, suggesting that fatigue during the pandemic involves alternative pathways and requires a multifactorial explanation.

Different underlying pathologic conditions have been described concerning postacute COVID-19 syndrome, including immune system dysfunction, mitochondrial dysfunction and vascular abnormalities, impaired oxygen consumption, as well as neuroinflammation, reactivation of herpesviruses or emergence of dysautonomia.⁶⁵ Many of these abnormalities and dysfunctions are observed in individuals with myalgic encephalomyelitis/chronic fatigue syndrome as well, suggesting an overlap in underlying mechanisms between these conditions.

Strengths and limitations

Study strengths include fatigue measurement into the COVID-19 pandemic, with a validated scale in a large randomly selected sample from the general population, and an assessment of various potential factors associated with fatigue. Results contributed to a better assessment of the population to date and confirmed some of the already existing risk factors that potentially constitute at-risk populations. Several limitations, however, need to be considered. First, despite random selection from state registers, and efforts to foster participation (study website providing plain-language scientific content, newsletters and large-audience webinars), participation bias emerged, particularly among vulnerable populations (those with low incomes, limited education and non-Swiss residents), older frail adults and young adults. However, additional standardisation of study sample did not alter our results. Second, there was a non-response bias for the general-health questionnaire with significant statistical differences between respondent and non-respondent groups. Major differences were observed in age groups (high participation by older adults and weak participation of young adults). Adjusting and weighting for the different factors helped mitigate these differences.

Conclusion and practical implications

About one-third of the adult general population of Geneva experienced severe fatigue, 2 years into the COVID-19 pandemic. Heightened fatigue among young adults is partly explained by depressive symptoms. Recent COVID-19 infection is substantially associated with severe fatigue, regardless of infection severity or co-occurrence of major depressive disorder. This study provides an updated fatigue level in the general population, and a baseline measurement to compare population-specific groups. Fatigue should be monitored over time using homogeneous and validated scales. Assessment in primary care settings by healthcare professionals for at-risk population groups could help identify and target modifiable risk factors.

Data availability statement

All data relevant to the study are included in the article or uploaded as supplementary information.