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

René G English¹ , Max O Bachmann² , Eric D Bateman¹ , Merrick F Zwarenstein³ , Lara R Fairall¹ , Angeni Bheekie⁴ , Bosielo P Majara⁵ , Carl Lombard⁶ , Robert Scherpbier⁷ and Salah Eddine Ottomani⁷

¹Knowledge Translation Unit, University of Cape Town Lung Institute, George Street, Mowbray, 7700, Cape Town, South Africa

²Department of Health Policy and Practice, School of Medicine, University of East Anglia, Norwich, NR47TJ, UK

³Knowledge Translation Programme and Department of Health Policy, Management and Evaluation, Faculty of Medicine, University of Toronto, Ontaria, M56IV7, Canada

⁴School of Pharmacy, University of the Western Cape, Bellville, 7535, Cape Town, South Africa

⁵Department of Community Health, University of the Free State, Bloemfontein, South Africa

⁶Biostatistics Unit, Medical Research Council, Tygerberg, Cape Town, South Africa

⁷Stop TB, World Health Organization, Geneva, Switzerland

author email corresponding author email

BMC Pulmonary Medicine 2006, 6:22doi:10.1186/1471-2466-6-22


Published:	25 August 2006

Abstract

Background

To evaluate the diagnostic accuracy of the integrated Practical Approach to Lung Health in South Africa (PALSA) guideline in identifying patients requiring bacteriological screening for tuberculosis (TB), and to determine which clinical features best predict suspected and bacteriologically-confirmed tuberculosis among patients with respiratory symptoms.

Methods

A prospective, cross-sectional study in which 1392 adult patients with cough and/or difficult breathing, attending a primary care facility in Cape Town, South Africa, were evaluated by a nurse using the guideline. The accuracy of a nurse using the guideline to identify TB suspects was compared to that of primary care physicians' diagnoses of (1) suspected TB, and (2) proven TB supported by clinical information and chest radiographs.

Results

The nurse using the guideline identified 516 patients as TB suspects compared with 365 by the primary care physicians, representing a sensitivity of 76% (95% confidence interval (CI) 71%–79%), specificity of 77% (95% CI 74%–79%), positive predictive value of 53% (95% CI 49%–58%), negative predictive value of 90% (95% CI 88%–92%), and area under the receiver operating characteristic curve (ARUC) of 0.76 (95% CI 0.74–0.79). Sputum results were obtained in 320 of the 365 primary care physicians TB suspects (88%); 40 (13%) of these were positive for TB. Only 4 cases were not identified by the nurse using the guideline. The primary care physicians diagnostic accuracy in diagnosing bacteriologically-confirmed TB (n = 320) was as follows: sensitivity 90% (95% CI 76%–97%), specificity 65% (95% CI 63%–68%), negative predictive value 7% (95% CI 5%–10%), positive predictive value 99.5% (95% CI 98.8%–99.8%), and ARUC 0.78 (95% CI 0.73–0.82). Weight loss, pleuritic pain, and night sweats were independently associated with the diagnosis of bacteriologically-confirmed tuberculosis (positive likelihood ratio if all three present = 16.7, 95% CI 5.9–29.4).

Conclusion

The PALSA guideline is an effective screening tool for identifying patients requiring bacteriological screening for pulmonary tuberculosis in this primary care setting. This supports the randomized trial finding that use of the guideline increased TB case detection.