Prevalence of severe fatigue after SARS-CoV-2 infection in Norway: a prospective 2-year follow-up study

Abstract

Objectives

This study aims to estimate the prevalence of long-lasting severe fatigue and identify possible risk factors in a 2-year follow-up of patients with predominantly mild-to-moderate SARS-CoV-2 infection.

Design

Prospective cohort study.

Setting

A community-based cohort from Telemark and Agder Counties, Norway.

Participants

A total of 159 PCR-confirmed SARS-CoV-2 positive individuals in the period between 28 February and 17 December 2020 were included at 12 months after SARS-CoV-2 infection, and 93 responded at 24 months follow-up.

Outcome measures

Fatigue was assessed using the Fatigue Severity Scale (FSS), and health-related quality of life using the RAND version of health-related quality of life Short Form 36 (SF-36), developed by the RAND Corporation. SARS-CoV-2 antibodies were measured at 12 and 24 months.

Results

Severe fatigue (FSS ≥5) was reported by 36% at 12 months and 31% at 24 months. A higher proportion of women than men reported severe fatigue at 12 months (p=0.08). The number of acute-phase symptoms was associated with severe fatigue. No association was found between severe fatigue and anti-SARS-CoV-2 antibody levels, demographic variables or reinfection status. The severe fatigue group scored significantly lower on all domains of SF-36.

Conclusion

In this cohort, severe fatigue was common, greatly impacted quality of life and persisted for up to 2 years following SARS-CoV-2 infection. Fatigue severity was associated with symptom burden in the acute phase but not with antibody levels or other demographic variables. These findings underscore the need for long-term follow-up and support for affected individuals.

STRENGTHS AND LIMITATIONS OF THIS STUDY.

• This study used a multicentre prospective design with a 2-year follow-up period, providing long-term data on post-COVID-19 fatigue.

• Fatigue and quality of life were assessed using well-validated tools (Fatigue Severity Scale and RAND Short Form 36).

• The cohort consisted primarily of non-hospitalised individuals, thereby increasing the generalisability of the findings.

• High loss to follow-up from baseline to 24 months may have introduced selection bias.

Introduction

Severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) has affected hundreds of millions of people worldwide, causing primarily acute respiratory infections, as well as symptoms from other organ systems. After the acute phase of SARS-CoV-2 infection, a substantial proportion of patients report long-lasting persistence of symptoms, a condition referred to as long covid, post-COVID-19 syndrome/condition or persistence of symptoms.¹ The WHO defines long covid as ‘continuation or development of new symptoms three months after the initial SARS-CoV-2 infection, with these symptoms lasting for at least 2 months with no other explanation’. Patients with long covid often have multiple symptoms, including fatigue, diminished exercise capacity, dyspnoea, depression, endocrine abnormalities, loss of taste or smell and diarrhoea,² among which fatigue is the most prevalent reported symptom.3,6

The prevalence of long covid substantially varies among studies.⁷ A systematic review and meta-analysis from the European Centre for Disease Prevention and Control published in October 2022 indicated a 51% overall prevalence of any post-COVID-19 condition symptoms (95% CI 41 to 60).⁸ Many of the included studies lacked SARS-CoV-2 negative control groups, thus potentially explaining the high prevalence. The study designs, measurements and cohorts showed heterogeneity across studies. A large observational cohort study from the Netherlands with matched controls and data from before SARS-CoV-2 infection has found a prevalence of 13% of persistent symptoms attributable to SARS-CoV-2.⁹ The mechanisms underlying post-COVID-19 condition have been extensively studied but remain incompletely understood and are likely multifactorial and complex.¹⁰ Several therapeutic approaches have been investigated, including interventions focusing on physical activity^{11 12} and rehabilitation programmes based on cognitive and behavioural principles,¹³ with some evidence of effectiveness. Although physical activity interventions are generally considered safe, they should be individually tailored, as inappropriate exercise may worsen long covid symptoms, such as post-exertional malaise.^{11 12}

Increased knowledge regarding the long-term outcomes of SARS-CoV-2 infected patients is needed. The use of validated scoring tools for objective and comparable assessments of reports is important. We performed a long-term follow-up study of a cohort of participants with SARS-CoV-2 PCR positivity. Our aims were to estimate the prevalence of severe fatigue 12 and 24 months after SARS-CoV-2 infection, and to identify possible risk factors in a 2-year follow-up of patients with predominantly mild to moderate SARS-CoV-2 infection. We hypothesised that severe fatigue could be associated with initial symptom burden, sex, comorbidities and antibody levels.

Methods

Study design and participants

This study is a follow-up study of the participants in the COVID-19 Telemark and Agder study (COVITA project), a multicentre cohort study on PCR confirmed SARS-CoV-2 infected patients and negative controls. In the COVITA project, all SARS-CoV-2 PCR positive adults aged ≥18 years residing in South-Eastern Norway (Agder and Telemark Counties) during the inclusion period were screened for eligibility, regardless of symptom status. The participants in the COVITA cohort were recruited by telephone from all hospitals in the region, municipality laboratories and COVID-19 test centres between 28 February and 17 December 2020. Exclusion criteria were age <18 years or inability to answer a questionnaire in Norwegian. All participants provided written informed consent before inclusion. The SARS-CoV-2 infected patients generally had mild/moderate disease, and only 6% were hospitalised. Detailed descriptions of the COVITA recruitment process, the follow-up of risk factors for SARS-CoV-2 infection and antibody persistence have previously been published.14,16 The study was authorised by the Regional Committee for Medical and Health Research Ethics of Southeast Norway (reference number 146469), the Norwegian Centre for Research Data (NSD) and the local Data Protection Officers in Telemark and Agder County.

In the present study, all participants with PCR-confirmed SARS-CoV-2 infection from the COVITA cohort were invited to reply to an online patient-reported outcome measure (PROM) questionnaire at 12 and 24 months. The invitation was sent by text message (SMS). Non-participation at follow-up reflects non-response to the SMS invitation. It was not possible to distinguish between active refusal and failure to receive or read the message. The participants were also invited to perform blood sampling at 12 and 24 months after PCR confirmed SARS-CoV-2. The 12-month serum samples were collected before SARS-CoV-2 vaccination for most participants.

Patient-reported outcome measures and other variables

PROMs are used for evaluating patients’ own perceptions of their health and quality of life. The 12-month PROM questionnaire included several validated measures: the Fatigue Severity Scale (FSS), the RAND version of the Medical Outcome Study 36-item Short Form (SF-36), developed by the RAND Corporation, and the Hospital Anxiety and Depression Scale (HADS). The FSS was included at both 12 and 24 months of follow-up. SF-36 and HADS were only included at 12 months follow-up.

Data on background variables were collected at inclusion. A summed comorbidity variable comprised the number of following conditions: asthma, chronic obstructive pulmonary disease, other chronic lung disease, cancer, heart disease, hypertension, diabetes, musculoskeletal disease and any other disease. The total number of initial symptoms at the first SARS-CoV-2 infection was based on the following symptoms: cough, runny nose, stuffy nose, sore throat, pain on swallowing, dyspnoea, headache, fever, fever with chills or sweating, abdominal pain, nausea or diarrhoea, impaired sense of smell or taste, myalgia and dizziness.

To address the lack of knowledge regarding the participants’ pre-SARS-CoV-2 health status, we asked the following question at 12 months follow-up: ‘How are you feeling currently on a scale from 10% to 100%, where 100% represents your health condition before SARS-CoV-2 infection’.

Fatigue Severity Scale

The FSS has been widely used for evaluating patients’ self-reported fatigue in various settings. Participants indicate their level of agreement with nine statements, on a scale from 1 (strongly disagree) to 7 (strongly agree), with increased scores indicating more severe fatigue. In 2022 Naik et al validated the FSS for fatigue measurement in a post-COVID-19 population.¹⁷ The cut-off for clinically relevant fatigue is usually set to ≥4 in the international literature¹⁸; however, according to a survey of the FSS in the general Norwegian population in 2005,¹⁹ most Norwegian studies use a cut-off ≥5 to avoid overdiagnosis of fatigue.

Medical Outcome Study SF-36.

The SF-36²⁰ is a frequently used tool to assess health-related quality of life. We used the RAND version translated into Norwegian. Population-based surveys for collecting normative data have been performed several times.²¹ The 36 items are grouped into eight domains: physical functioning, role limitations due to physical problems, bodily pain, general health, vitality, social functioning, role limitations due to emotional problems and mental health. The scores were calculated using a scale from 0 to 100, with scores in the lower range representing poorer health status. The calculations were performed according to the SF-36 scale syntax provided by the Norwegian public health institute.²² Finally, the physical and mental component summary scores were calculated.

Hospital Anxiety and Depression Scale

The HADS has been widely used in clinical practice and research to measure anxiety and depression. HADS total (HADS-T) consists of 14 questions, subdivided into two scales: anxiety (HADS-A) and depression (HADS-D), consisting of 7 questions each. The scale has been evaluated in a Norwegian population²³ and shown to have good overall psychometric properties for psychological and emotional stress assessment in patients with long covid.²⁴ Scores ≥8 for HADS-A and HADS-D, and ≥11 for HADS-T are commonly used cut-offs indicating the need for further assessment.^{23 25}

Laboratory methods

The serum samples were prepared from whole blood centrifuged for 10 min at a minimum of 1800 g and stored at −80°C until further analysis. All antibody analyses were performed on a Cobas e 801 instrument (Roche Diagnostics). For quantitative measurement of SARS-CoV-2 spike antibody, an Elecsys Anti-SARS-CoV-2 S electrochemiluminescence immunoassay from Roche was used. This assay quantifies total antibodies by using an S antigen recombinant receptor binding domain protein in a double-antigen sandwich assay. All testing and interpretation were conducted according to the manufacturer’s instructions. Because of the very high spike antibody concentrations at 24 months, samples with >250 units/millilitre (U/mL) were diluted several times and re-measured. The following dilutions were used: 1:10, 1:50 and 1:400, thus providing a measurement range of 0.4–100 000 U/mL.

For qualitative detection of SARS-CoV-2 nucleocapsid antibodies, the Elecsys Anti-SARS-CoV-2 electrochemiluminescence immunoassay from Roche was used. Nucleocapsid antibodies were analysed and interpreted according to the manufacturer’s instructions, wherein a Cut-Off-Index (COI) <1 was considered negative.

Statistics

Statistical analyses were performed in IBM SPSS Statistics V.29.0.0. Variables are presented as means with SD, medians with IQR and absolute frequencies, as relevant. Categorical variables were compared with the χ² test, whereas continuous variables were compared with the independent t-test. Univariable and multivariable logistic regression were performed with generalised linear models. Variables were entered simultaneously for the multivariable analysis. Pearson’s correlation was used for comparison of the continuous variables FSS versus mental and physical component summary scores. P values<0.05 were considered statistically significant. We did not perform any imputation for missing data. For variables with incomplete data, we reported the number of participants with available information for each variable in the results tables. Analyses were conducted using available data for each parameter. The proportion of missing data was low and unlikely to affect the results. We compared selected baseline characteristics between participants who completed follow-up and those lost to follow-up. Analyses at 24 months were based on available data, without imputation. Due to heavily skewed distributions, the serum levels of SARS-CoV-2 nucleocapsid and spike antibodies were categorised into high and low levels, according to a 50th percentile cut-off.

Patient and public involvement

Two user representatives of SARS-CoV-2 PCR-positive patients were involved in the establishment of the COVITA project. They gave feedback on the study protocol, study methods, information and consent forms and questionnaires.

Results

A total of 656 individuals with PCR-confirmed SARS-CoV-2 infection were screened for eligibility. Of these, 391 participants, mainly non-hospitalised, were included in the COVITA cohort and invited to take part in this study (figure 1). A total of 159 participants (41%) (80 women and 79 men) answered the 12-month PROM questionnaire and were included. The mean age was 46.8 (SD 14.4) and 53.7 (SD 11.6) years for women and men, respectively. Serum samples were collected from 140 participants at 12 months. At 24 months, 93 (59%) participants completed the questionnaire, and 126 provided a serum sample. Among these participants, 85 provided both a serum sample and completed the questionnaire at 24 months. The population characteristics are listed in table 1. The lost to follow-up group at 24 months (n=66) was a little younger (48 vs 53 years), had a higher number of initial symptoms at the first SARS-CoV-2 infection (6.7 vs 5.6) and a higher score on HADS total (11.1 vs 8.7). There was no difference between the groups regarding sex, comorbidities, antibody level or the presence of severe fatigue at 12 months.

Figure 1. Included participants. FSS, Fatigue Severity Scale; HADS, Hospital Anxiety and Depression Scale; PROM, patient-reported outcome measure; RAND 36, RAND version of the Medical Outcome Study 36-item Short Form (SF-36).

Table 1. Population characteristics.

Characteristics	All participants (N=159)
Age, mean (SD)	50.3 (±13.5)
Sex
Male	79 (50%)
Female	80 (50%)
BMI, mean (SD)	26.6 (±4.4)
Education N=156
Primary and secondary	15 (10%)
High school and certificate	54 (35%)
University, 4 years or fewer	60 (38%)
University, more than 4 years	27 (17%)
Smoking N=155
Non-smoker	86 (56%)
Past smoker	52 (33%)
Daily and occasional smoker	17 (31%)
Asthma N=156	23 (15%)
COPD N=152	2 (1%)
High blood pressure N=156	18 (12%)
Diabetes N=156	8 (5%)
Comorbidity sum*, N=152
0 comorbidities	89 (59%)
≥1 comorbidities	63 (41%)
Number of initial symptoms†, N=152, mean (SD)	6.1 (±3.1)
HADS-total N=159, mean (SD)	9.7 (±7.7)
HADS depression	3.9 (±4)
HADS anxiety	5.8 (±4.4)
HADS-total ≥11, N (%)	61 (38)
HADS depression score ≥8, N (%)	29 (18)
HADS anxiety score ≥8, N (%)	53 (33)

^*Comorbidity sum: number of patients with ≥1 of the following conditions: asthma, COPD, other chronic lung disease, cancer, heart disease, hypertension, diabetes, musculoskeletal disease and any other disease.

^†Number of the following initial symptoms at the first SARS-CoV-2 infection: cough, runny nose, stuffy nose, sore throat, pain on swallowing, dyspnoea, headache, fever, fever with chills or sweating, abdominal pain, nausea or diarrhoea, impaired sense of smell or taste, myalgia and dizziness.

BMI, body mass index; COPD, chronic obstructive pulmonary disease; HADS, Hospital Anxiety and Depression Scale.

Fatigue Severity Scale

A FSS score ≥5 was categorised as severe fatigue. At 12 months after SARS-CoV-2 infection, 36% of participants (57/159) reported severe fatigue. After 24 months, 10 participants showed a decrease in the fatigue score from severe to normal, whereas 9 participants showed an increase from normal to severe. The proportion of participants with severe fatigue at both 12 and 24 months was approximately one-third (figure 2).

Figure 2. Prevalence of severe fatigue at 12 and 24 months. FSS, Fatigue Severity Scale.

Of the 30 participants with FSS scores ≥5 who also answered the follow-up questionnaire after 24 months, 21 (70%) reported decreased fatigue (lower mean FSS score of 1.1 points, SD 0.99).

Of the 93 participants who answered the 24-month follow-up questionnaire, 31 (33%) reported a second episode of SARS-CoV-2 infection during the study period. SARS-CoV-2 reinfection was not correlated with worsened FSS scores at 24 months (p=0.52).

The FSS demonstrated excellent internal consistency (Cronbach’s α=0.97 (12 months) and 0.98 (24 months)).

In the response to the question ‘How are you feeling currently on a scale from 10% to 100%, where 100% represents your health condition before SARS-CoV-2 infection?’ the group with FSS scores ≥5 had significantly lower scores than the group with FSS scores <5; the mean values were 61% (±23) and 88% (±14), respectively (p<0.001) (figure 3).

Figure 3. Distribution of answers, categorised by FSS ≥5 or FSS <5. No participants with severe fatigue indicated a health condition 100% of their pre-SARS-CoV-2 infection health status at 12 months. FSS, Fatigue Severity Scale.

The association between anti-SARS-CoV-2 antibodies and fatigue.

The cut-offs for low and high SARS-CoV-2 antibody levels at 12 and 24 months were 13.6 COI and 30.9 COI for anti-nucleocapsid antibody, and 210 U/mL and 9510.50 U/mL for anti-spike antibody. Our data indicated no association between antibody levels and severe fatigue at 12 or 24 months. The results are shown in online supplemental S1. At 24 months, all participants had been vaccinated, and 91% had received two or more doses. Figures showing the median antibody levels, illustrating the lack of difference between the groups with severe fatigue versus those without are available in online supplemental S2.

Quality of life

At the 12-month survey, all participants completed the SF-36 evaluating health-related quality of life.

Table 2 compares the SF-36 domains between participants with severe fatigue and those with normal FSS scores. Participants with severe fatigue scored significantly lower in all domains.

Table 2. SF-36 scores in all participants, grouped by FSS ≥5 versus FSS <5.

SF-36 domain	All participants N=159	FSS ≥5 N=57 (SD)	FSS <5 N=102 (SD)	P value*
Physical function	85.0 (±19.4)	73.3 (±20.2)	91.5 (±15.5)	<0.001
Role physical	63.1 (±41.1)	27.63 (±32.6)	82.8 (±30.9)	<0.001
Body pain	72.6 (±24.8)	56.1 (±22.8)	81.9 (±20.8)	<0.001
General health	65.1 (±22.6)	47.7 (± 17.9)	74.8 (±18.9)	<0.001
Vitality	54.0 (±25.8)	30.1 (±17.7)	67.4 (±19.1)	<0.001
Social functioning	77.0 (±23.7)	57.0 (±22.0)	88.1 (±16.2)	<0.001
Role emotional	66.9 (±41.5)	38.0 (± 41.5)	83.0 (±31.7)	<0.001
Mental health	74.4 (±18.6)	61.0 (±18.0)	81.8 (±14.2)	<0.001

^*P value calculated for comparison between FSS ≥5 versus FSS <5.

FSS, Fatigue Severity Scale; SF-36, 36-item Short Form.

A plot of FSS scores as a continuous variable indicated that higher FSS scores correlated with lower SF-36 scores in both the mental component summary and physical component summary (p<0001). Figures available in online supplemental S3. Internal consistency was good to excellent, Cronbach’s alpha values for the eight domains ranged from 0.86 (general health) to 0.93 (vitality).

Predictors of severe fatigue

Background variables were compared between groups to identify severe fatigue predictors or risk factors at 12 months (table 3). The number of acute symptoms during the first episode of SARS-CoV-2 was associated with severe fatigue. Participants with severe fatigue also scored significantly higher on both HADS anxiety and depression than participants with normal scores. Women were more prevalent than men in the severe fatigue group (p=0.078). In multivariable logistic regression, only the HADS-T score was statistically significant (table 4).

Table 3. Comparison of background variables between participants with fatigue severity scale ≥5 and <5 at 12 months.

Characteristic	FSS ≥5 N=57	FSS <5 N=102	P value
Age, mean (SD)	49.5 (±14.6)	50.7 (±12.9)	0.59
Sex			0.078
Male	23 (40.4%)	56 (54.9%)
Female	34 (59.6%)	46 (45.1%)
BMI, mean (SD)	26.5 (±4.0)	26.7 (±4.6)	0.82
Education N=156			0.35
Primary and secondary	5 (9.1%)	10 (9.9%)
High school and certificate	24 (43.6%)	30 (29.7%)
University, 4 years or fewer	19 (34.5%)	41 (40.6%)
University, more than 4 years	7 (12.7%)	20 (19.8%)
Smoking N=155			0.65
Non-smoker	27 (49%)	59 (59%)
Past smoker	22 (40%)	30 (30%)
Daily and occasional smoker	6 (11%)	11 (11%)
Asthma N=156	11 (20.0%)	12 (11.9%)	0.17
COPD N=152	1 (1.9%)	1 (1.0%)	0.67
High blood pressure N=156	4 (7.3%)	14 (13.9%)	0.22
Diabetes N=156	1 (1.8%)	7 (6.9%)	0.17
Comorbidity sum*, N=152			0.68
0 comorbidities	31 (56%)	58 (60%)
≥1 comorbidities	24 (44%)	39 (40%)
Number of initial symptoms†, N=152, mean (SD)	6.9 (±3.0)	5.6 (±3.0)	0.012§
HADS-T‡ N=159, mean (SD)	14.9 (±7.9)	6.8 (±5.9)	<0.001§
HADS depression	6.3 (±4.2)	2.6 (±3.2)	<0.001§
HADS anxiety	8.61 (±4.6)	4.2 (±3.3)	<0.001§
HADS-T ≥11, N (%)	37 (65)	24 (24)	<0.001§
HADS depression ≥8, N (%)	18 (32)	11 (11)	0.001§
HADS anxiety ≥8, N (%)	36 (63)	17 (17)	<0.001

^*Comorbidity sum: Number of patients with the following conditions: asthma, COPD, other chronic lung disease, cancer, heart disease, hypertension, diabetes, musculoskeletal disease and any other disease.

^†Number of the following initial symptoms at the first SARS-CoV-2 infection: Cough, runny nose, stuffy nose, sore throat, pain on swallowing, dyspnoea, headache, fever, fever with chills or sweating, abdominal pain, nausea or diarrhoea, impaired sense of smell or taste, myalgia and dizziness.

^‡HADS-T: The Hospital Anxiety and Depression Scale total score.

^§Significant results are indicated with an asterisk.

BMI, body mass index; COPD, chronic obstructive pulmonary disease; FSS, Fatigue Severity Scale.

Table 4. Univariable and multivariable logistic regression models of possible predictors for severe fatigue at 12 months.

Variable	Univariable analysis OR (95% CI)	P value	Multivariable analysis OR (95% CI)	P value
Sex Female Male	Ref 0.6 (0.3 to 1.2)	0.12	Ref 0.5 (0.2 to 1.2)	0.12
NC* antibodies, 12 months
Low	Ref		Ref
High	1.0 (0.5 to 2.1)	0.95	1.4 (0.5 to 3.7)	0.51
Spike antibodies, 12 months
Low	Ref		Ref
High	1.4 (0.7 to 2.9)	0.34	1.2 (0.4 to 3.0)	0.77
HADS-T†	1.2 (1.1 to 1.3)	<0.001	1.2 (1.1 to 1.3)	<0.001
Symptom score‡	1.2 (1.1 to 1.4)	0.005	§
Comorbidity score¶
0 comorbidities	Ref		Ref
≥1 comorbidities	1.3 (0.6 to 2.7)	0.45	1.3 (0.6 to 2.9)	0.58

^*Nucleocapsid antibody.

^†HADS-T: The Hospital Anxiety and Depression Scale total score.

^‡Number of the following initial symptoms at the first SARS-CoV-2 infection: cough, runny nose, stuffy nose, sore throat, pain on swallowing, dyspnoea, headache, fever, fever with chills or sweating, abdominal pain, nausea or diarrhoea, impaired sense of smell or taste, myalgia and dizziness.

^§Due to collinearity between symptom score and HADS-T, symptom score was not included in the multivariable model.

^¶Comorbidity score: number of the following conditions present: asthma, COPD, other chronic lung disease, cancer, heart disease, hypertension, diabetes, musculoskeletal disease and any other disease.

COPD, chronic obstructive pulmonary disease.

Discussion

Prevalence of fatigue and quality of life

We assessed the 2-year outcome in mainly non-hospitalised SARS-CoV-2 infected patients during the first and second waves of the COVID-19 pandemic. Approximately one-third of the patients had severe fatigue at both the 12-month and 24-month follow-up, but most patients reported improvement of symptoms at the 24-month follow-up. Health-related quality of life was significantly lower in the group with severe fatigue than in the group with normal FSS scores.

The prevalence of fatigue in our cohort was lower than reported by most other studies. Seeßle et al reported fatigue in approximately 50% of patients 12 months after SARS-CoV-2 infection.²⁶ That study included 96 patients, 32% of whom were hospitalised. In a Swedish cohort of 433 hospitalised SARS-CoV-2 patients, 40% experienced persistent symptoms after 4 months and among these, 84% still reported symptoms after 2 years.²⁷ Fatigue was among the most reported symptoms, at a rate of 65%. The lower prevalence of fatigue in our population may be due to our cohort consisting of mainly non-hospitalised patients, different assessment tools, or a stricter definition of severe fatigue. Another long-term follow-up study by Peghin et al reported a rate of post-COVID-19-related symptoms comparable to our study. In a cohort of 230 patients, 36% were categorised as having post-COVID-19 syndrome after 2 years.²⁸ That cohort was a combination of inpatients and outpatients, and the main symptom was fatigue. None of the long-term follow-up studies described above used the FSS. Although we did not conduct a complete assessment of all possible post-COVID-19 symptoms in our study, we demonstrated a significant correlation between FSS scores ≥5 and a poorer health condition after than before SARS-CoV-2 infection. We therefore presume that, for most patients in our cohort, fatigue started after the SARS-CoV-2 infection.

Although the rate of severe fatigue was similar at 12 and 24 months in our cohort, 70% of patients with severe fatigue at 12 months had lower FSS scores at 24 months than at 12 months, indicating a favourable prognosis over time. Other long-term follow-up studies have also reported improvements in symptoms, although the total burden of issues remains high.²⁷

Patients with severe fatigue scored significantly lower in all eight domains of SF-36 than patients without severe fatigue. AlRasheed et al followed a large group of previously SARS-CoV-2 infected individuals (n=400) and non-infected controls (n=565) for up to 24 months.²⁹ Infected individuals had significantly lower scores than controls in only three SF-36 domains: role limitations due to physical problems, vitality and role limitations due to emotional problems. The choice of different groups for comparison in those studies may explain these differences. Our finding of a negative correlation between the FSS score and both mental and physical component summary scores was in accordance with the study by AlRasheed et al.²⁹ As fatigue is among the most prevalent symptoms of post-COVID-19 syndrome,^{4 5 26} and the fatigue score correlated well with the SF-36 score, the FSS might serve as a favourable and simple indicator of the syndrome.

Antibody levels as a predictor of severe fatigue

We found no correlation between anti-SARS-CoV-2 nucleocapsid or spike antibody levels and severe fatigue at 12 or 24 months. Peghin et al have reported elevated antibody levels in patients with post-COVID-19 syndrome,³⁰ and Hackenbruch et al have shown elevated SARS-CoV-2 antibody levels in patients with more than three post-infectious symptoms.³¹ Both those studies were performed 3–6 months after infection. However, other studies have shown that initially low SARS-CoV-2 antibodies correlate with severe fatigue,⁶ measured at 3 months.

Studies with longer follow-up times have reported conclusions in line with our findings. Seeßle et al found no correlation between SARS-CoV-2 antibody levels and prolonged symptoms after 12 months.²⁶ Peghin et al, who reported higher antibody levels at 6 months in patients with post-COVID-19 syndrome, did not observe any correlation with antibodies and long covid after 2 years in the same population.²⁸ A study among UK healthcare workers infected with SARS-CoV-2 with a 12-month follow-up did not demonstrate differences in antibody levels.³ Based on our data and findings from other studies, differences in SARS-CoV-2 antibody response are unlikely to impact the prevalence of post-COVID-19 syndrome.

Predictors of severe fatigue

A Norwegian study by Stavem et al, determining risk factors for fatigue after SARS-CoV-2 infection, found that female sex and the number of acute COVID-19 symptoms were associated with fatigue risk.³² An increased risk for females and the association of the number of acute COVID-19 symptoms to post-COVID-19 symptoms has also been confirmed by others.^{26 28} Our data also indicate a significant association between higher numbers of initial COVID-19 symptoms and severe fatigue. The proportion of women was larger than men in the group with severe fatigue, but the difference was not statistically significant. Unlike other studies,²⁸ our study indicates no correlation between the number of comorbidities and fatigue, possibly because the majority (59%) of our participants reported no comorbidities.

The group with severe fatigue scored significantly higher on HADS, in both univariable and multivariable analysis, than the group with normal FSS scores. Other studies have reported a relatively high prevalence of anxiety and depression symptoms in patients after COVID-19, measured with HADS. Hussain et al have observed anxiety symptoms in 38% and depression symptoms in 35% of patients after intensive care unit admission for COVID-19 at the 1-year follow-up.³³ Similar results have been confirmed by others after hospitalisation for COVID-19.³⁴ Based on our data, we cannot assume that psychological and emotional stress, measured with HADS, is a risk factor for severe fatigue after SARS-CoV-2 infection. The association might also be due to long covid symptoms leading to increased anxiety and depression symptoms. It is also important to acknowledge that individuals with long covid who experience social vulnerabilities may be more likely to develop symptoms of anxiety and depression.³⁵ The causality of the association between anxiety and depression symptoms and long covid warrants further study.

Strengths and limitations

The strengths of this study include its multicentre prospective design and the long follow-up period of 2 years. The assessments of fatigue and quality of life were conducted with well-validated tools. The population consisted primarily of non-hospitalised participants, thus supporting generalisability to the general population, given that most SARS-CoV-2 infected individuals had mild disease and were treated as outpatients.

Study limitations include the absence of a SARS-CoV-2 negative control group, lack of linkage to the Norwegian Cause of Death Registry and missing information on the severity of SARS-CoV-2 reinfections. The response rate at 2 years was relatively low. The loss-to-follow-up analysis showed that factors associated with an increased risk of severe fatigue in our cohort were more prevalent in the group lost to follow-up. This suggests that the low response rate is unlikely to have introduced bias leading to an overestimation of fatigue at the 2-year follow-up. Another limitation is that we lack information on the included participants’ fatigue status and quality of life before the SARS-CoV-2 infection. We addressed this aspect by asking the participants how they rated their current health condition compared with their pre-COVID-19 condition.

Conclusion

Severe fatigue was common among our participants after SARS-CoV-2 infection, greatly impacted quality of life and could persist for more than 24 months. The number of symptoms in the acute phase of SARS-CoV-2 infection was associated with severe fatigue. This study did not demonstrate any correlation between severe fatigue and SARS-CoV-2 antibody levels.

Data availability statement

Data are available upon reasonable request.