Skip to main content

Table 1 Baseline characteristics

From: Symptom-based scoring technique by machine learning to predict COVID-19: a validation study

Variable

Category – metric

Setting

Referral center

(n = 555)

Emergency department

(n = 199)

p-valuea

Non-missing

Missing

Outcome

 COVID-19 by RT-PCR test

Negative – n (%)

203 (36.58)

9 (4.52)

 > .05

 < .001

Positive – n (%)

327 (58.92)

10 (5.03)

Missing – n (%)

25 (4.50)

180 (90.45)

 

Index test

 COVID-19 by symptom scoring b

Negative – n (%)

263 (47.39)

197 (99.00)

 > .05

 < .001

Positive – n (%)

35 (6.31)

0 (0.00)

Missing – n (%)

257 (46.30)

2 (1.00)

 

Comparator tests

 COVID-19 by antigen test

Negative – n (%)

19 (3.42)

30 (15.08)

 < .001

.004

Positive – n (%)

29 (5.23)

2 (1.00)

Missing – n (%)

507 (91.35)

167 (83.92)

 

 COVID-19 by antibody test

Negative – n (%)

87 (15.68)

5 (2.51)

 > .05

 < .001

Positive – n (%)

31 (5.58)

0 (0.00)

Missing – n (%)

437 (78.74)

194 (97.49)

 

Predictors

 Age (years)

Non-missing – mean ± 95% CI

42.91 ± 1.74

50.25 ± 2.26

 < .001

 > .05

Missing – n (%)

8 (1.44)

0 (0.00)

 

 Sex

Female – n (%)

279 (50.27)

90 (45.23)

 > .05

 > .05

Male – n (%)

268 (48.29)

109 (54.77)

Missing – n (%)

8 (1.44)

0 (0.00)

 

 Loss of smell

No – n (%)

318 (57.30)

199 (100)

 > .05

 > .05

Yes – n (%)

58 (10.45)

0 (0.00)

Missing – n (%)

179 (32.25)

0 (0.00)

 

 Loss of taste

No – n (%)

295 (53.15)

198 (99.50)

.031

 > .05

Yes – n (%)

14 (2.52)

1 (0.50)

Missing – n (%)

246 (44.33)

0 (0.00)

 

 Cough

No – n (%)

158 (28.47)

183 (91.96)

 < .001

 > .05

Yes – n (%)

367 (66.13)

16 (8.04)

Missing – n (%)

30 (5.40)

0 (0.00)

 

 Fatigue

No – n (%)

212 (38.20)

139 (69.85)

.015

 > .05

Yes – n (%)

145 (26.13)

60 (30.15)

Missing – n (%)

198 (35.67)

0 (0.00)

 

 Skipped meals

No – n (%)

274 (49.37)

174 (87.44)

.048

 < .001

Yes – n (%)

61 (10.99)

23 (11.56)

Missing – n (%)

220 (39.64)

2 (1.00)

 
  1. a, Complete p-value indicates the statistical significance of the between-setting difference in data distribution using only non-missing values, while the missing p-value indicates the statistical significance of the between-setting difference in missing proportion; b only samples without missing values in any predictors under the default threshold (i.e., 0.5). CI Confidence interval, NA Not applicable, RT-PCR Reverse-transcription polymerase chain reaction