Skip to main content
. 2022 Aug 4;17(8):e0271310. doi: 10.1371/journal.pone.0271310

Table 4. Associations of the number and pattern (latent-class analysis) of acute COVID-19 symptoms with any chronic COVID-19 symptom, particularly chronic fatigue, anosmia and dysgeusia.

Variable ≥1 chronic COVID-19 symptom
No Yes Crude OR (95% CI) P-value Adjusted OR (95% CI) P-value
Freq (%) Freq (%)
Number of acute COVID-19 symptoms
    0–2 56 (90.3%) 6 (9.7%) 1.000 (ref) 1.000 (ref)
    3–5 157 (74.4%) 54 (25.6%) 3.21 (1.31–7.87) 0.011 3.22 (1.31–7.95) 0.011
    ≥6 67 (67.7%) 32 (32.3%) 4.46 (1.74–11.42) 0.002 4.23 (1.64–10.93) 0.003
Pattern of acute COVID-19 symptoms (latent-class analysis)
    1. Highest probability of all symptoms 32 (64.0%) 18 (36.0%) 3.70 (1.38–9.88) 0.009 3.62 (1.33–9.83) 0.012
    2. Lowest probability of all symptoms 46 (86.8%) 7 (13.2%) 1.000 (ref) 1.00 (ref)
    3. Fever, cough, muscle ache, anosmia, dysgeusia, headache 135 (72.6%) 51 (27.4%) 2.48 (1.05–5.85) 0.038 2.32 (0.97–5.55) 0.059
    4. Fever, cough, muscle ache, headache 68 (80.0%) 17 (20.0%) 1.64 (0.63–4.28) 0.309 1.63 (0.65–4.02) 0.322
  Chronic fatigue
No Yes Crude OR (95% CI) P-value Adjusted OR (95% CI) P-value
Freq (%) Freq (%)
Number of acute COVID-19 symptoms
    0–2 67 (98.5%) 1 (1.5%) 1.00 (ref) 1.00 (ref)
    3–5 196 (89.1%) 24 (10.9%) 8.20 (1.09–61.82) 0.041 8.48 (1.12–64.17) 0.038
    ≥6 82 (81.2%) 19 (18.8%) 15.52 (2.03–118.99) 0.008 14.62 (1.90–112.56) 0.01
Pattern of acute COVID-19 symptoms (latent-class analysis)
     1. Highest probability of all symptoms 40 (78.4%) 11 (21.6%) 7.84 (1.65–37.29) 0.01 7.36 (1.53–35.47) 0.013
    2. Lowest probability of all symptoms 57 (96.6%) 2 (3.4%) 1.00 (ref) 1.00 (ref)
    3. Fever, cough, muscle ache, anosmia, dysgeusia, headache 170 (88.1%) 23 (11.9%) 3.86 (0.88–16.87) 0.073 3.58 (0.81–15.85) 0.094
    4. Fever, cough, muscle ache, headache 80 (90.9%) 8 (9.1%) 2.85 (0.58–13.92) 0.196 2.84 (0.58–13.96) 0.2
  Chronic anosmia
No Yes Crude OR P-value Adjusted OR P-value
Freq (%) Freq (%) (95% CI) (95% CI)
Number of acute COVID-19 symptoms
    0–2 65 (95.6%) 3 (4.4%) 1.00 (ref) 1.000 (ref)
    3–5 195 (88.6%) 25 (11.4%) 2.78 (0.81–9.50) 0.226 2.81 (0.82–9.70) 0.102
    ≥6 92 (91.1%) 9 (8.9%) 2.12 (0.55–8.13) 0.247 2.01 (0.52–7.77) 0.313
Pattern of acute COVID-19 symptoms (latent-class analysis)
     1. Highest probability of all symptoms 46 (90.2%) 5 (9.8%) 1.17 (0.32–4.31) 0.809 1.00 (0.27–3.71) 0.997
    2. Lowest probability of all symptoms 54 (91.5%) 5 (8.5%) 1.00 (ref) 1.00 (ref)
    3. Fever, cough, muscle ache, anosmia, dysgeusia, headache 167 (86.5%) 26 (13.5%) 1.68 (0.61–4.59) 0.311 1.47 (0.53–4.08) 0.462
    4. Fever, cough, muscle ache, headache 86 (97.7%) 2 (2.3%) 0.25 (0.05–1.34) 0.106 0.24 (0.04–1.27) 0.092
  Chronic dysgeusia
No Yes Crude OR P-value Adjusted OR P-value
Freq (%) Freq (%) (95% CI) (95% CI)
Number of acute COVID-19 symptoms
    0–2 67 (98.5%) 1 (1.5%) 1.00 (ref) 1.00 (ref)
    3–5 202 (91.8%) 18 (8.2%) 5.97 (0.78–45.57) 0.085 6.03 (0.79–46.31) 0.084
    ≥6 98 (97.0%) 3 (3.0%) 2.05 (0.21–20.14) 0.538 1.93 (0.20–19.04) 0.574
Pattern of acute COVID-19 symptoms (latent-class analysis)
    1. Highest probability of all symptoms 48 (94.1%) 3 (5.9%) 1.17 (0.23–6.05) 0.854 0.97 (0.19–5.12) 0.975
    2. Lowest probability of all symptoms 56 (94.9%) 3 (5.1%) 1.00 (ref) 1.00 (ref)
    3. Fever, cough, muscle ache, anosmia, dysgeusia, headache 177 (91.7%) 16 (8.3%) 1.69 (0.47–6.00) 0.419 1.49 (0.41–5.40) 0.545
    4. Fever, cough, muscle ache, headache 87 (98.9%) 1 (1.1%) 0.22 (0.02–2.12) 0.187 0.20 (0.02–1.95) 0.165

Bivariable logistic regression models were constructed with ≥1 chronic COVID-19 symptom or chronic fatigue, anosmia and dysgeusia (yes vs. no) as the dependent variable and number of acute COVID-19 symptoms (0-2/3-5/≥6) and pattern of acute COVID-19 symptoms (ordinal variable with 4-classes derived from latent-class analysis) as the independent variable. Analyses were limited to persons with positive SARS-CoV-2 anti-IgG antibodies. Crude odds ratios (OR) and 95% confidence intervals (CI) were estimated. Multivariable models included age (continuous), sex (male/female), race (white/non-white) and Hispanic ethnicity (yes/no) as covariables, and state of residence. Adjusted OR and 95% CI were estimated.

Bold-face indicates statistical significance (P<0.05).