Skip to main content

Table 2 Diagnostic accuracy at various time windows

From: Performance of the ROX index in predicting high flow nasal cannula failure in COVID-19 patients: a systematic review and meta-analysis

Time window

Number of studies

Number of patients

Number of cut-offs

sAUC

Optimal cut-off

Sensitivity at optimal cut-off

Specificity at optimal cut-off

Cut-off for 80% sensitivity

Cut-off for 80% specificity

p-value*

≥ 2 h, < 6 h

5

743

6

0.754 (95% CI: 0.604–0.863)

5.71

0.635 (95% CI: 0.411–0.813)

0.769 (95% CI: 0.465–0.927)

7.96

5.33

NA

≥ 6 h, < 12 h

8

1126

10

0.795 (95% CI: 0.287–0.908)

6.50

0.835 (95% CI: 0.389–0.976)

0.659 (95% CI: 0.192–0.940)

5.95

3.69

0.069

≥ 12 h, < 24 h

7

2452

10

0.821 (95% CI: 0.713–0.894)

5.78

0.749 (95% CI: 0.575–0.868)

0.759 (95% CI: 0.511–0.905)

6.07

5.55

0.001

  1. sAUC area under the summary receiver operator characteristic curve; CI confidence interval; NA not applicable
  2. *p-value for difference, using ≥ 2 h, < 6 h window as the base comparator
Back to article page

Critical Care

ISSN: 1364-8535