Use and evaluation of a line probe assay in patients with tuberculosis in Peru: 2011-2013.

Objective To determine the use and performance of a line probe assay (LPA) compared with conventional culture and drug sensitivity testing (CDST) in patients registered with tuberculosis (TB) under routine program conditions in Peru in 2011-2013. Methods This was a descriptive, operational research, cross-sectional study of sputum specimens from patients with smear-positive pulmonary TB and mycobacterial cultures from patients with smear-negative or positive TB. Drug resistance to rifampicin and/or isoniazid detected by LPA was compared to CDST. Sensitivity, specificity, and predictive values were calculated and reliability for detecting drug resistance was assessed through kappa coefficient, with values 0.61-0.80 showing substantial correlation, and 0.81 or above showing almost-perfect correlation. Results In 2011-2013, there were 16 169 LPA tests performed, with the proportion of TB patients receiving the test increasing from 3.2% to 30.2%. In all, 2 905 LPA test results were compared to CDST. For LPA in sputum specimens, sensitivity for rifampicin was 92%; isoniazid, 94%; and MDR-TB, 88%; while specificity for rifampicin was 92%; isoniazid, 92%; and MDR-TB, 95%. For LPA in mycobacterial cultures, sensitivity for rifampicin was 95%; isoniazid, 96%; and MDR-TB, 90%; while specificity for rifampicin was 85%; isoniazid, 91%; and MDR-TB, 94%. Kappa coefficients were at 0.81 or above for all comparisons of LPA with CDST using sputum specimens and cultures, except for isoniazid in cultures, which was at 0.79. Conclusions This study suggests that LPA is a reliable and rapid screening test for drug-resistant TB and should be considered suitable for routine use and scale up in Peru.