Skip to main content
. 2024 Jun 11;79(1):161–168. doi: 10.1093/cid/ciae239

Table 1.

Sensitivity, Specificity, and Diagnostic Odds Ratio of Polymerase Chain Reaction (PCR) Test According to PCR Technique

Sample Sensitivity
(95% CI)
Specificity
(95% CI)
DOR
(95% CI)
LR+
(95% CI)
LR
(95% CI)
BALF samples
 qPCR (n = 2673) 0.987
(.968–.995)
0.893
(.844–.927)
635
(269–1498)
9.194
(5.727–12.661)
0.014
(.001–.027)
 cPCR (n = 2254) 0.972
(.932–.988)
0.954
(.930–.970)
710
(305–1652)
21.178
(12.438–29.918)
0.30
(.004–.056)
IS samples
 qPCR (n = 491) 0.980
(.944–.993)
0.815
(.721–.883)
217
(78–601)
5.303
(3.024–7.583)
0.024
(.000–.049)
 cPCR (n = 590) 0.956
(.887–.984)
0.917
(.866–.950)
243
(90–656)
11.511
(5.985–17.036)
0.047
(.001–.094)
URT samples
 qPCR (n = 352) 0.892
(.710–.965)
0.905
(.809–.955)
78
(26–238)
9.340
(2.997–15.682)
0.120
(NE–.245)
 cPCR (n = 512) 0.787
(.502–.931)
0.960
(.911–.982)
87
(27–284)
19.424
(5.358–33.490)
0.222
(NE–.446)

Sensitivity, specificity, DORs, and positive and negative likelihood ratios with 95% CI calculated from the coefficients of the binomial regression model. In regard to human immunodeficiency virus (HIV) versus non-HIV status, there was no significant difference in the cohorts of patients tested by cPCR or qPCR or by specimen types tested. Numbers of samples (n) where analysis of sensitivity/specificity was performed by sample type are shown. “NE” (not evaluable) appears when the corresponding results obtained by the calculating algorithm had a negative sign, which is impossible as the LRs are ratios between 2 positive numbers (proportions). This fact is due to the use of an approximate method to calculate the 95% CIs of the LRs, the “delta method.”

Abbreviations: BALF, bronchoalveolar lavage fluid; CI, confidence interval; cPCR, conventional polymerase chain reaction; DOR, diagnostic odds ratio; IS, induced sputum; LR+, positive likelihood ratio; LR, negative likelihood ratio; NE, not evaluable; qPCR, quantitative polymerase chain reaction; URT, upper respiratory tract.