Somatosensory processing in long COVID fatigue and its relations with physiological and psychological factors

Figures

• 

  FIGURE 1. Concept of interactions. Schematic diagram indicating the proposed relations between autonomic dysfunction, illness beliefs, mood (anxiety and depression) and somatosensory processing. We postulate that each of these factors (autonomic dysfunction, illness beliefs, mood) will impact fatigue via an impact on somatosensory processing. We also propose that fatigability and acute (state) changes in fatigue will have a bidirectional relation with somatosensory processing.

• 

  FIGURE 2. Data collection timeline. JISC is a digital platform designed specifically for education and research organisations, that can be used for the creation and dissemination of surveys. BL, baseline; PRE‐1, immediately after pre‐task somatosensory attenuation measure; PRE‐2, immediately after pre‐task somatosensory gating measure; POST‐1, immediately after the completion of the task (cognitive or physical); POST‐2, immediately after post‐task somatosensory attenuation measure; POST‐3, immediately after post‐task somatosensory gating measure; HRV, heart rate variability.

• 

  FIGURE 3. Distribution across measures used to evaluate fatigue severity (a), impact of fatigue on functioning (b), mood (c), autonomic symptoms (c), interoceptive sensibility (d, e) and illness perceptions (f) for control (blue) and long COVID (orange) participants who attended at least one in‐person testing session. Sample sizes vary by measure due to missing data (control n = 42–44; long COVID n = 43–44; see Table 3 for details). ‘n’ represents the number of independent participants contributing to each measure. The y‐axes are scaled to show the full range of possible scores on each measure. Boxes show the median and interquartile range; whiskers extend to the most extreme data points within 1.5× the interquartile range. Individual data points are shown by crosses to the left of each box plot. Black dashed lines show clinical severity cutoffs: FAS < 22 = normal, 22–34 = mild‐to‐moderate, ≥35 = severe; CFQ‐11 bimodal ≥4 = excessive fatigue; MFIS ≥38 = high impact; COMPASS‐31 0–20 = mild, 21–100 = moderate‐severe; B‐IPQ < 42 = low, 42–49 = moderate, ≥50 = high threat. The equivalent plot for all participants who completed the online questionnaires is provided in Supplementary material 7,…

• 

  FIGURE 4. Number of days between consecutive testing sessions for control (blue)and long COVID (orange) participants. Boxes show the median and interquartile range; whiskers extend to the most extreme data points within 1.5× the interquartile range. Individual data points are shown by crosses to the left of each box plot. Sample sizes vary due to some participants only completing one session (long COVID: n = 38; control: n = 41). ‘n’ represents the number of independent participants contributing to each measure.

• 

  FIGURE 5. Visual analogue scale (VAS) score for fatigue at each of six time points throughout the cognitive (VAS; a) and physical (VAS; b) testing session for control (blue) and long COVID (orange) participants. Thin lines represent individual data points. Thick lines represent median values. The VAS scale was 100 mm long and anchored on the left side (0 mm) with ‘not fatigued at all’ and on the right side (100 mm) with ‘extremely fatigued’. VAS score for fatigue was evaluated at baseline (BL), immediately after pre‐task somatosensory attenuation measure (PRE‐1), immediately after pre‐task somatosensory gating measure (PRE‐2), immediately after the completion of the task (cognitive or physical) (POST‐1), immediately after post‐task somatosensory attenuation measure (POST‐2) and immediately after post‐task somatosensory gating measure (POST‐3). Sample sizes were consistent across time points for controls (cognitive: n = 42; physical: n = 43) but varied slightly for long COVID participants (cognitive: n = 41–42; physical: n = 38–40) due to missing data and incomplete sessions. ‘n’ represents the number of independent participants contributing to each measure.

• 

  FIGURE 6. Heart rate and heart rate variability measured during quiet sitting in the pre‐task assessment of sensory attenuation. Heart rate (HR; a, c) and heart rate variability (HRV; b, d) measured during quiet sitting in the pre‐task assessment of sensory attenuation in the cognitive (a, b) and physical (c, d) testing sessions for control (blue) and long COVID (orange) participants. Boxes show the median and interquartile range; whiskers extend to the most extreme data points within 1.5× the interquartile range. Individual data points are shown by crosses to the left of each box plot. HRV was calculated as percentage of successive inter‐beat intervals that differed by more than 50 ms. Sample sizes vary across the cognitive (long COVID: n = 26; control: n = 26) and physical sessions (long COVID: n = 24; control: n = 24) due to missing data and incomplete sessions. ‘n’ represents the number of independent participants contributing to each measure. Mann–Whitney U‐test P‐values for group comparisons are displayed within the figure.

• 

  FIGURE 7. Fatigue during cognitive (top row) and physical (bottom row) tasks. Left panels (a, c) show numerical rating scale (NRS) score at each of four time points during the cognitive task (a) and the physical task (c) for control (blue) and long COVID (orange) participants. Right panels (b, d) show change in NRS score from the first to the fourth time point during the cognitive task (b) and the physical task (d). Boxes show the median and interquartile range; whiskers extend to the most extreme data points within 1.5× the interquartile range. Individual data points are shown by crosses to the left of each box plot. Time points (T1–T4) indicate when NRS ratings were collected: after every fourth task and at the end of the cognitive session, and at 90 s, 180 s, 270 s and at the end of the 6‐min walk test for the physical task. Sample sizes vary across the cognitive (long COVID: n = 42; control: n = 42) and physical sessions (long COVID: n = 40; control: n = 43) due to missing data and incomplete sessions. ‘n’ represents the number of independent participants contributing to each measure. Mann–Whitney U‐test P‐values for group comparisons are displayed within the figure.

• 

  FIGURE 8. Perceived exertion during cognitive (a, b) and physical (c, d) tasks. Left panels (a, c) show Borg rating of perceived exertion score at each of four time points during the cognitive task (a) and the physical task (c) for control (blue) and long COVID (orange) participants. Right panels (b, d) show change in Borg score from the first to the fourth time point during the cognitive task (b) and the physical task (d). Boxes show the median and interquartile range; whiskers extend to the most extreme data points within 1.5× the interquartile range. Individual data points are shown by crosses to the left of each box plot. Time points (T1–T4) indicate when Borg ratings were collected: after every fourth task and at the end of the cognitive session, and at 90 s, 180 s, 270 s and at the end of the 6‐min walk test for the physical task. Sample sizes vary across the cognitive (long COVID: n = 42; control: n = 42) and physical sessions (long COVID: n = 40; control: n = 43) due to missing data and incomplete sessions. ‘n’ represents the number of independent participants contributing to each measure. Mann–Whitney U‐test P‐values for group comparisons are displayed within the figure.

• 

  FIGURE 9. Task performance during the cognitive (a, b) and physical (c, d) tasks. Left panels (a, c) show performance at the start and end of the cognitive task (a) and the physical task (c) for control (blue) and long COVID (orange) participants. Right panels (b, d) show change in performance (fatigability) across the task during the cognitive task (ΔBIS) (b) and the physical task (Δ walking speed) (d). Boxes show the median and interquartile range; whiskers extend to the most extreme data points within 1.5× the interquartile range. Individual data points are shown by crosses to the left of each box plot. For the cognitive task, performance was evaluated using the balanced integration score (BIS) in the AX‐CPT task at the start and end of the cognitive testing battery. For the physical task, performance was evaluated using average walking speed at the start (30–60 s) and end (330–360 s) of the 6‐min walk test. Sample sizes were consistent across time points for the cognitive task (long COVID: n = 42; controls: n = 42) but varied slightly for the physical session (long COVID: n = 37–39; control: n = 36–42) due to missing data. ‘n’ represents the number of independent participants…

• 

  FIGURE 10. Heart rate (HR) during cognitive (a, b) and physical (c, d) tasks. Left panels (a, c) show HR at each of four time points during the cognitive task (a) and the physical task (c) for control (blue) and long COVID (orange) participants. Right panels (b, d) show change in HR from the first to the fourth time point during the cognitive task (b) and the physical task (d). Boxes show the median and interquartile range; whiskers extend to the most extreme data points within 1.5× the interquartile range. Individual data points are shown by crosses to the left of each box plot. Sample sizes vary across the cognitive (long COVID: n = 20; control: n = 20) and physical sessions (long COVID: n = 12; control: n = 12) due to missing data and incomplete sessions. ‘n’ represents the number of independent participants contributing to each measure. Mann–Whitney U‐test P‐values for group comparisons are displayed within the figure.

• 

  FIGURE 11. Heart rate variability (HRV) during cognitive (a, b) and physical (c, d) tasks. Left panels (a, c) show HRV at each of four time points during the cognitive task (a) and the physical task (c) for control (blue) and long COVID (orange) participants. Right panels (b, d) show change in HRV from the first to the fourth time point during the cognitive task (b) and the physical task (d). Boxes show the median and interquartile range; whiskers extend to the most extreme data points within 1.5× the interquartile range. Individual data points are shown by crosses to the left of each box plot. Sample sizes vary across the cognitive (long COVID: n = 20; control: n = 20) and physical sessions (long COVID: n = 12; control: n = 12) due to missing data and incomplete sessions. ‘n’ represents the number of independent participants contributing to each measure. Mann–Whitney U‐test P‐values for group comparisons are displayed within the figure.

• 

  FIGURE 12. Somatosensory attenuation during the cognitive (a, b) and physical (c, d) testing sessions. Left panels (a, c) show somatosensory attenuation before and after the cognitive task (a) and the physical task (c) for control (blue) and long COVID (orange) participants. Right panels (b, d) show change in somatosensory attenuation from pre‐ to post‐cognitive (b) and physical (d) task. Boxes show the median and interquartile range; whiskers extend to the most extreme data points within 1.5× the interquartile range. Individual data points are shown by crosses to the left of each box plot. Sample sizes vary across both the cognitive (long COVID: n = 42; control: n = 42) and the physical sessions (long COVID: n = 39; control: n = 42) due to missing data and incomplete sessions. ‘n’ represents the number of independent participants contributing to each measure. Mann–Whitney U‐test P‐values for group comparisons are displayed within the figure.