RESUMEN
BACKGROUND AND OBJECTIVES: Patients with multidrug-resistant (MDR) tuberculosis (TB) are at high risk of treatment failure. It is anticipated that early identification of MDR-TB and appropriate treatment will improve patient outcome and disease control. We evaluated the rapid detection of rifampicin resistance in previously treated TB patients, directly from acidfast bacilli (AFB)-positive sputum using a phage-based test, FASTPlaque-Response (Biotec Laboratories Ltd, Ipswich, UK). The ability of rifampicin resistance to predict MDR-TB was also determined. DESIGN: A prospective study was done comparing performance of the rapid phage test with conventional culture and drug susceptibility testing (DST) in AFB-positive TB patients. SETTING: Five primary health clinics and one TB referral centre in the Port Elizabeth Metropolitan area, Eastern Cape. OUTCOME MEASURES: Sensitivity, specificity and overall accuracy of the phage test were determined compared with gold standard culture and DST. Discrepant results were resolved by molecular detection of mutations conferring rifampicin resistance. The proportion of rifampicin-resistant strains that were MDR was also determined. RESULTS: Previously treated patients were at a high risk of MDRTB (35.7%). Sensitivity, specificity and overall accuracy of FASTPlaque-Response for rifampicin resistance determination were 95.4% (95% confidence interval (CI): 91.0 - 99.8%), 97.2% (95% CI: 94.5 - 99.9%) and 96.5% (95% CI: 94.1 - 98.9%) respectively compared with conventional DST (unresolved), calculated for specimens that had both FASTPlaque-Response and conventional DST results available. FASTPlaque-Response results were available in 2 days instead of 28 - 85 days with conventional DST. However, only 70.6% of FASTPlaque-Response results were interpretable compared with 86.3% of conventional DST results. The majority (95.5%) of rifampicinresistant strains were MDR-TB. CONCLUSIONS: Rapid detection of rifampicin resistance using FASTPlaque-Response could contribute to improved management of patients at risk of MDR-TB, such as previously treated patients. However, improvement in control of specimen-related contamination is needed to ensure that a higher proportion of FASTPlaque-Response results are interpretable. Where indicated, early modification of therapy could improve patient prognosis and reduce disease transmission.