Subdomains of Post-COVID Syndrome (PCS) - a population-based study

Abstract

Purpose: 'Post-COVID Syndrome' (PCS), which encompasses the multifaceted sequelae of COVID-19, can be severity-graded by a previously defined score encompassing 12 different long-term symptom complexes. The PCS score was shown to have two main predictors, namely acute COVID-19 severity and individual resilience. The purpose of the present study was to verify these predictors and to assess their detailed relationship to the symptom complexes constituting the PCS score.

Methods: The study drew upon a largely expanded dataset (n = 3,372) from COVIDOM, the cohort study underlying the original PCS score definition. Classification and Regression Tree (CART) analysis served to resolve the detailed relationship between the predictors and the constituting symptom complexes of the PCS score.

Results: Among newly recruited COVIDOM participants (n = 1,930), the PCS score was again found to be associated with both its putative predictors. Of the score-constituting symptom complexes, neurological symptoms, sleep disturbance, and fatigue were predicted by individual resilience whereas the acute disease severity predicted exercise intolerance, chemosensory deficits, joint or muscle pain, signs of infection, and fatigue. These associations inspired the definition of two novel PCS scores that included the above-mentioned subsets of symptom complexes only. Similar to the original PCS score, both novel scores were found to be inversely correlated with quality of life as measured by the EQ-5D-5L index.

Conclusion: The two newly defined PCS scores may enable a more refined assessment of PCS severity, both in a research context and to delineate distinct PCS subdomains with possibly different therapeutic and interventional needs in clinical practise.

Keywords: COVID-19; Cohort study; Coronavirus; Disease severity; Long-COVID; Phenotype classification; Post-COVID syndrome; Resilience.

Fig. 1

Relationship between PCS score and individual resilience (a) and acute COVID-19 severity (b) in newly recruited COVIDOM participants with complete data (resilience: n = 770; acute severity: n = 1,268). Individual resilience was classified as low (BRS < 3.0), medium (3.0 ≤ BRS < 4.3), or strong (BRS ≥ 4.3). Acute COVID-19 severity was classified as none (no severe or life-threatening symptoms), weak (1–3 symptoms), moderate (4–6 symptoms), or severe (≥ 7 symptoms). Both omnibus and all pairwise tests revealed significant score differences between the predictor-defined groups (Kruskal-Wallis test or paired Wilcoxon test, as appropriate; ****: Benjamini-Hochberg-adjusted p < 0.001)

Fig. 2

Pair-wise associations between the 12 long-term symptom complexes underlying the Post-COVID Syndrome (PCS) score. Cohen’s κ values were calculated in all COVIDOM participants with complete data (n = 2,889) and subjected to hierarchical clustering by Euclidean distance. The strength of the association between symptom complexes was color-coded as red (0.75 ≥ k), yellow (0.45 ≤ κ < 0.75) or blue (κ < 0.45)

Fig. 3

Classification and Regression Tree (CART) analysis of individual resilience, as measured by the Brief Resilience Scale (BRS) index. CART analysis served to determine which PCS symptom complexes were most strongly associated with individual resilience in all COVIDOM participants with complete data (n = 2,250), allowing for a possible statistical interaction between complexes. a Nodes were successively split by the presence (1, blue) or absence (0, white) of a particular symptom complex. The percentage of samples included is given alongside each node. The resulting tree was pruned until all leaves comprised at least 5% of the samples. b Symptom complexes were selected for novel PCS score definition by applying the elbow rule to their importance values (threshold marked by red triangle)

Fig. 4

Classification and Regression Tree (CART) analysis of acute COVID-19 severity (n = 2,889; for details, see legend to Fig. 3)