Within and between-day variation and associations of symptoms in Long Covid: Intensive longitudinal study 2023 Burton et al

Abstract

Background
People with Long Covid (Post Covid-19 Condition) describe multiple symptoms which vary between and within individuals over relatively short time intervals. We aimed to describe the real-time associations between different symptoms and between symptoms and physical activity at the individual patient level.

Methods and findings
Intensive longitudinal study of 82 adults with self-reported Long Covid (median duration 12–18 months). Data collection involved a smartphone app with 5 daily entries over 14 days and continuous wearing of a wrist accelerometer. Data items included 7 symptoms (Visual Analog Scales) and perceived demands in the preceding period (Likert scales). Activity was measured using mean acceleration in the 3-hour periods preceding and following app data entry. Analysis used within-person correlations of symptoms pairs and both pooled and individual symptom networks derived from graphical vector autoregression. App data was suitable for analysis from 74 participants (90%) comprising 4022 entries representing 77.6% of possible entries. Symptoms varied substantially within individuals and were only weakly autocorrelated. The strongest between-subject symptom correlations were of fatigue with pain (partial coefficient 0.5) and cognitive difficulty with light-headedness (0.41). Pooled within-subject correlations showed fatigue correlated with cognitive difficulty (partial coefficient 0.2) pain (0.19) breathlessness (0.15) and light-headedness (0.12) but not anxiety. Cognitive difficulty was correlated with anxiety and light-headedness (partial coefficients 0.16 and 0.17). Individual participant correlation heatmaps and symptom networks showed no clear patterns indicative of distinct phenotypes. Symptoms, including fatigue, were inconsistently correlated with prior or subsequent physical activity: this may reflect adjustment of activity in response to symptoms. Delayed worsening of symptoms after the highest activity peak was observed in 7 participants.

Conclusion
Symptoms of Long Covid vary within individuals over short time scales, with heterogenous patterns of symptom correlation. The findings are compatible with altered central symptom processing as an additional factor in Long Covid.

Open access, https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0280343

 

"Interpretation

Four aspects of our findings point to a potential role of disordered interoception and symptom processing in Long Covid. These are (1) the within person variability of symptoms; (2) the relationship between symptoms and activity, (3) the between person variation in patterns of symptoms; (4) the cluster of light-headedness, breathlessness, and cognitive difficulty.

The marked within-person variability indicates that symptoms vary within and between days in ways which appear largely unpredictable and inconsistently related to activity or demand. While this variation could be explained by highly varying pathophysiology, it may also be explained by disturbed processing of interoceptive signals such that symptoms represent inaccurate markers of the body’s state. The weak association of symptoms, including fatigue, with activity is similar to that seen in musculoskeletal [50] and neurological disease [51]. This does not invalidate patients’ experience of fatigue and other symptoms, rather it emphasises its complexity. In suggesting the possibility that altered interoception may play a role in Long Covid we are not arguing that other mechanisms are not present; rather our data suggests that, on their own, peripheral mechanisms may not fully account for the patterns of symptoms or their severity. Indeed, peripheral pathophysiology and central processes such as interoception are complementary and may be causally linked both through changes in brain function due to vascular and inflammatory processes underpinning changes in interoceptive processing and through the close links between interoception and autonomic signalling [23]. One cluster of associations that did feature in our data was of light-headedness, cognitive difficulty, and breathlessness. These may relate to changes in areas of the brain involved in interoceptive processing [28] or in autonomic regulation [3, 53]. Dysfunctional breathing [19, 54] which might account for some of this cluster can also be understood using an interoception framework

Together these findings can be viewed from the perspective of an embodied predictive interoceptive coding model of symptoms [24, 56]. Indeed, viewing physical symptoms as altered body signals from the body through an impaired interoceptive system may be analogous to the widely recognised parosmia observed during recovery of taste and smell after covid [57]. Addressing Long Covid from this perspective has the potential to be both non-stigmatising and to suggest additional therapeutic approaches and treatments. Several interventions appear capable of improving interoception: these include vagal nerve stimulation and behavioural techniques such as slow paced breathing [58]. Some of these are currently being provided in interventions for Long Covid and all warrant further study. Further research should use methods similar to those described here, with more closely defined groups of patients and in parallel with investigations of other pathophysiological mechanisms or within trials of interventions.

Conclusion

Symptoms of Long Covid vary within individuals over short time scales, with heterogenous patterns of symptom correlation and weak associations with concurrent or preceding physical activity. These findings are in keeping with an embodied predictive interoceptive coding model of symptoms, suggesting that Long Covid is associated with changes to the way the brain processes signals from the body."

 

This looks like a study with some effort and thought behind it. However the interpretation that this is a problem of symptom processing seems a little dubious. It seems just a variation of "we don't understand it, therefore it's psychological" with different language being used.

I would have expected a clearer association between activity levels and symptom exacerbation. Maybe the relatively short study duration is hiding some "cumulative PEM", or maybe mental activity is adding symptom exacerbations that appear indepent of exertion because the mental activity wasn't being measured? Or maybe the patients were mostly pacing fairly well and not having major exacerbations because of that?

 

To put things in perspective, my last bad next-day crash was last summer because I avoid the kind of activity level that is likely to cause them if possible. I've had many smaller next-day day crashes in that time, and also many crashes from cumulative exertion over several days or weeks. Or so I interpret events. Maybe they aren't seeing much association between symptoms and activity because they're assuming that patients are crashing hard all the time, within 3 days, and have done a statistical analysis based on this assumption.

 

So exactly as the patients have reported. For decades. A basic fact that is disputed and reattributed instead to some BS speculative fantasy.

What is this weird bit of speculation in the damn abstract? It shouldn't even be in the paper.

The whole symptoms as a signal thing is ridiculous. Even in normal conversation when people talk about something being a symptom of something, it clearly means an impairment, not a signal. It's not a red light incorrectly reporting that something is clogged up, it's something being clogged up, or broken, it means something can't be used as intended. The symptoms are the impairment, any other framing is just nonsense. The whole interpretation section has no place in a scientific paper, it's pure opinion that makes no sense of the evidence, doesn't bother with validity.

Chronic illness is often fluctuating and relapsing-remitting. This has been known and reported for decades, and it is still disputed in favor of baseless speculation. Also as noted by Trish: 2 weeks is not longitudinal, this is not serious and definitely not intensive. It's been 3 years and it's well-known, or easy to find out, that many have been sick for the whole duration.

Huge problem in interpretation with the weak association with activity: it's not a linear pattern. This kind of superficial data could only notice linear predictable patterns. It's not linear, therefore you don't get a clear pattern, because the pattern of the illness is unpredictable. Enough of this nonsense, medicine should deal with the disease they have in front of them, not the one they wished they had.

 

I have not read the article yet, but with ME variation in symptoms does not just relate to previous physical or cognitive activity levels, but also needs to take into account orthostatic issues, sensory hypersensitivities and diet in those with food intolerances.

Also given PEM is a potential state change that also increases sensitivity to over exertion, but also increases the effect of the other triggering factors such as the sensory hypersensitivity you would not expect to see a simple linear relationship.

Certainly, personally I only began to make sense of variation in my ME when I considered not only physical or cognitive exertion, but also things like the amount of time I spent standing, levels of light and background sound and what I ate recently.

 

Nice demo of PEM/PESE in figure 10: https://journals.plos.org/plosone/article/figure?id=10.1371/journal.pone.0280343.g010

But only in 10% of participants. But I guess that in itself is useful?

Over 2/3rds had had LC for more than a year, so presumably most had learnt to pace by then anyway, if they weren't already improving?