Skip to main content

Table 1 Computation of the relative specificity to exclude cervical intra-epithelial neoplasia of grade 2 or worse of Selfy on self-samples (SS) vs clinician-taken samples (clin) and non-inferiority statistics

From: Can HPV Selfy be considered as a clinically validated HPV test for use in cervical cancer screening?

Correct statistic

 
 

Selfy clin−

Selfy clin+

 

Selfy SS−

708

16

724

Selfy SS+

37

30

67

 

745

46

791

Specificity Selfy SS

 = 724/791 = 

91.5%

 

Specificity Selfy clin

 = 745/791 = 

94.2%

 

Relative specificity SS/clin

0.97

(95% CI 0.95–0.99)*

 

T non inferiority

− 0.86

  

p non-inferiority

0.81

  

Wrong statistic (b and c cells switched in the abcd matrix)

 

Selfy clin−

Selfy clin+

Selfy SS−

708

37

745

Selfy SS+

16

30

46

 

724

67

791

Specificity Selfy clin

 = 745/791 = 

94.2%

 

Specificity Selfy SS

 = 724/791 = 

91.5%

 

Relative specificity clin/SS

1.03

(95% CI 1.01–1.05)*

T non inferiority

6.60

  

p non-inferiority

< 0.0001

  
  1. In italics: non-inferiority statistic reported by Avian et al. [1] which was due to erroneous switching the values 37 and 16. In fact this reported statistic reflects that Selfy on clin samples is not inferior to SS samples