Risk factors for fatigue severity in the post-COVID-19 condition: A prospective controlled cohort study of nonhospitalised adolescents and young adults

Abstract

Long COVID is a global health concern, leading to persistent symptoms and disability long after the acute SARS-CoV-2 infection in most age groups. The condition can manifest even following mild COVID-19, and in young people, it may have serious adverse consequences for educational attainment and transition to adulthood. Fatigue is the most prevalent symptom, but the underlying mechanisms remain poorly understood. In this prospective study of 404 SARS-CoV-2-positive and 105 SARS-CoV-2 negative, non-hospitalised youth (ages 12-25, female 62%), we investigated which factors in the early convalescent stage (<28 days since test) were associated with the severity of persistent fatigue at 6 months after infection. Participants completed questionnaires regarding clinical symptoms, social factors and psychological traits, and were subject to clinical and functional testing and biomarker analyses. Variables with significant (p < 0.2) associations to the outcome in simple linear regression were chosen for multivariable modelling, together with potential confounders. In the final multivariable model, SARS-CoV-2-positivity was a minor risk factor for fatigue severity at six months. Baseline severity of symptoms was the main risk factor and correlated with psychosocial factors such as loneliness and neuroticism, rather than biomarkers. Our results suggest that factors not related to infection are major risk factors for persistent fatigue in this age group.

Keywords: Adolescents; Chronic fatigue syndrome; Long COVID; Post-COVID-19 condition; Post-acute sequelae of COVID-19; Post-infective fatigue syndrome; SARS-CoV-2.

Fig. 1

Trajectories of fatigue severity from baseline to six months, n = 467 Caption: Fatigue severity assessed by the Chalder fatigue scale (Chalder et al., 1993), range 0–33, where higher scores imply more severe fatigue. Three participants had missing fatigue scores at baseline.

Fig. 2

Baseline risk factors and their univariate associations to fatigue severity^a∗. Caption: Linear regression. Only associations with p < 0.5 are shown for brevity. Values for all variables are given in Supplemental table S5. Footnotes: BMI=Body mass index; CI = 95% Confidence interval; SARS-CoV-2= Severe acute respiratory syndrome coronavirus 2; FVC=Forced vital capacity; IL=Interleukin. ^aAs measured by the Chalder Fatigue Questionnaire, score 0–33, higher scores imply more fatigue. ^bFrom the Hopkins Verbal Learning Test revised (HVLT-R); higher scores imply better delayed recall of words. ^cFrom the DePaul Symptom Questionnaire; higher score implies more frequent post-exertional malaise. ^dFrom the Karolinska Sleep Questionnaire; higher score implies better sleep. ^eFrom the Brief Pain Inventory, higher score implies more pain. ^fSelf-developed, aggregated score for problems with ‘memory’, ‘concentration’, and ‘decision making’; higher score implies more symptoms. ^gSelf-developed, aggregated score for symptoms ‘cough’ and ‘dyspnoea’; higher score implies more symptoms. ^hSelf-developed, aggregated score for symptoms ‘dizziness’, ‘cold and pale hands’, ‘feeling alternately warm and cold’; higher score implies more symptoms. ⁱFrom the anxiety and depression subscales, respectively, of the Hospital Anxiety and Depression Scale; higher scores imply more symptoms. ^jFrom the Positive and Negative Affect Schedule; higher score implies more negative emotions. ^kThe main component extracted by Principal Component Analysis of the 10 clinical symptoms variables, labelled ‘symptom severity’. ^lFrom the NEO-Five-Factor-Inventory-30; higher scores implies more neuroticism. ^mFrom the Toronto Alexithymia Scale; higher score implies more difficulty identifying feelings. ⁿFrom the Penn State Worry Questionnaire; higher score implies more worrying. ^oFrom the Body Vigilance Scale; higher score implies being more attentive to bodily sensations. ^pThe main component extracted by Principal Component Analysis of the four psychological traits variables, labelled ‘emotional maladjustment’. ^qSelf-developed; higher score implies more physical activity. ^rFrom the University of California, Los Angeles, Loneliness Scale; higher score implies more loneliness. ^oFrom the Life Event Checklist; higher score implies more negative impact of past life events. ∗ ln(x+1), † natural logarithm, ‡square-root, § Cube root, and ¶ fifth root transformations were applied to respective variables for regression analyses.