Positioning of the Xpert® MTB/RIF diagnostic system in a rural health system in KwaZulu-Natal, South Africa: preliminary findings from a cluster randomised trial

Lessells, R.J., Cooke, G.S., Nicol, M.P., McGrath, N., Newell, M.L. and Godfrey-Faussett, P. (2012) Positioning of the Xpert® MTB/RIF diagnostic system in a rural health system in KwaZulu-Natal, South Africa: preliminary findings from a cluster randomised trial. 43rd World Conference on Lung Health of the International Union against Tuberculosis and Lung Disease, Kuala Lumpur, Malaysia. 13 - 17 Nov 2012.

Abstract

Background:
The Xpert® MTB/RIF system is a molecular diagnostic for the detection of Mycobacterium tuberculosis and rifampicin resistance. Data on diagnostic performance and clinical outcomes at different levels of the health system are required to inform scale-up within high burden countries.

Methods:
A cluster randomised trial comparing two Xpert® MTB/RIF positioning strategies for adult pulmonary TB suspects: point-of-care [POC] strategy (system positioned at primary health care clinic) vs. hospital laboratory [HL] strategy (system positioned at district hospital laboratory). Unit of randomisation S270 Abstract presentations, Friday, 16 November is two-week time block; participants recruited from single clinic if: age ? 18 yrs, current cough, HIV infection ± high risk of MDR-TB. This preliminary analysis incorporates first twelve randomisation blocks (7 POC, 5 HL) and focuses on diagnostic performance and retention in diagnostic pathway. Sensitivity and specificity for detection of M. tuberculosis were calculated with reference to a single MGIT culture.

Results:
486 individuals enrolled; 446 submitted specimens (266 POC, 180 HL). Xpert positivity was similar in both strategies (17.7% POC vs. 13.9% HL, P=0.29). Sensitivity for detection of M. tuberculosis was 80.0% (95%CI 63.9–90.4) at POC and 71.4% (95%CI 42.0–90.4) at HL. Specificity was 96.1% (95%CI 91.3–98.4) at POC and 91.4% (95%CI 84.3–95.6) at HL. The proportion of tests with invalid/error results was similar (3.4% POC vs. 4.4% HL, P=0.62). The proportion of participants who received their result within 30 days was 96.6% POC vs. 79.4% HL (P<0.001).

Conclusions and recommendations:
These preliminary results suggest that implementation of the Xpert® MTB/RIF system at primary health care clinic level produces comparable diagnostic performance to implementation within the normal laboratory system but reduces loss to follow-up within the diagnostic process. The ongoing trial will explore the impact on individual clinical outcomes.

This record has no associated files available for download.

Contributors

Author: N. McGrath

Author: M.L. Newell

Download statistics