Table 4.
Multivariable analysis, effect size, and validation of models.
| Composite index/inflammatory marker | Model | Effect size* |
Validation |
|||
|---|---|---|---|---|---|---|
| OR (CI 95%) | p | AUC (CI 95%) | p | Standardized residuals | ||
| Mean ± SD | ||||||
| Women | ||||||
| C1 index | Logistic regression (DV: PCS) | Adjusted OR = 5.14 (1.6–16.4) | 0.006 | 0.76 (0.63–0.89) | 0.001 | – |
| C3 index | Logistic regression (DV: PCS) | Adjusted OR = 4.20 (1.3–13.3) | 0.015 | 0.72 (0.57–0.86) | 0.005 | – |
| C4 index | Logistic regression (DV: PCS) | Adjusted OR = 4.12 (1.3–13.1) | 0.016 | 0.71 (0.57–0.85) | 0.006 | – |
| Neutrophil count (×103/µL) | General linear model (Fixed factor: Anosmia) | (Anosmia+) 3.43 ± 0.3 (Anosmia−) 2.58 ± 0.1 | 0.014 | – | – | 0 ± 0.95 |
| Neutrophil count (×103/µL) | General linear model (Fixed factor: Ageusia) | (Ageusia+) 3.89 ± 0.3 (Ageusia−) 2.59 ± 0.1 | 0.002 | – | – | 0 ± 0.95 |
| Men | ||||||
| CRp > 0.3 and < 1.0 mg/dL | Logistic regression (DV: PCS) | Adjusted OR = 12.9 (1.3–121) | 0.025 | 0.76 (0.58–0.94) | 0.025 | – |
| C2 index | Logistic regression (DV: PCS) | Adjusted OR = 10.1 (1.2–85) | 0.033 | 0.75 (0.55–0.95) | 0.029 | – |
| C5 index | Logistic regression (DV: PCS) | Adjusted OR = 17.5 (2–153) | 0.010 | 0.82 (0.65–0.98) | 0.003 | – |
| Neutrophil count (×103/µL) | General linear model (Fixed factor: Fatigue) | (Fatigue+) 4.68 ± 0.6 (Fatigue−) 3.37 ± 0.1 | 0.041 | – | 0 ± 0.93 | |
* After adjusting for age, BMI, Charlson index and tobacco consumption.
Abbreviations. DV, Dependent variable; PCS, Post-COVID syndrome; CRP, C-reactive protein; BMI, Body mass index; OR, odds ratio; CI, confidence interval; AUC, Area under the (ROC) curve; SD, Standard deviation.
Validation of the models: AUC values ranged between 0.71 and 0.82, and the standardised residuals showed a normal distribution, with a curve of mean = 0 and SD = 1. R2 values were 0.27, 0.54, 0.23, 0.23, 0.58 and 0.54, for composite indices C1, C2, C3, C4 and C5, respectively.