Skip to main content

Table 5 Independent and combined likelihood ratios for predictors of suspected or diagnosed tuberculosis: logistic regression models

From: Diagnostic accuracy of an integrated respiratory guideline in identifying patients with respiratory symptoms requiring screening for pulmonary tuberculosis: a cross-sectional study

Reference standard

Predictors

Positive likelihood ratio

(95% CI)

Negative likelihood ratio

(95% CI)

Tuberculosis suspected by respiratory physician*

     
 

Previous tuberculosis

2.6

(1.9–3.5)

0.75

(0.67–0.84)

 

Haemoptysis

4.8

(2.2–10.3)

0.91

(0.86–0.96)

 

Night sweating

2.8

(1.8–4.2)

0.77

(0.68–0.87)

 

Weight loss

3.5

(2.3–5.6)

0.75

(0.66–0.84)

 

Pleuritic pain

2.5

(1.5–4.1)

0.90

(0.84–0.95)

 

No difficulty breathing

1.6

(1.2–2.1)

0.88

(0.80–0.96)

 

No wheeze

1.2

(1.1–1.4)

0.70

(0.53–0.92)

 

All 7 predictors

594

(232–1821)

0.22

(0.17–0.29)

Tuberculosis diagnosed by primary care physician**

     
 

Weight loss

2.9

(1.3–5.8)

0.81

(0.67–0.96)

 

Night sweats

2.0

(1.0–4.6)

0.76

(0.54–0.97)

 

Pleuritic pain

2.9

(1.0–4.8)

0.82

(0.63–1.00)

 

All 3 predictors

16.7

(5.9–29.4)

0.50

(0.36–0.65)

  1. * area under ROC curve for model = 0.85. ** area under ROC curve for model = 0.74