LAUSR

Background There is need for simple, cost effective and widely available point of care tests for low level health facilities in developing countries to screen for drug resistant tuberculosis (TB) after bacteriological confirmation of TB by smear microscopy. We evaluated the sensitivity and specificity of the mean corpuscular volume (MCV) and CD4/CD8 ratio in discriminating between rifampicin resistant (RR-TB) and rifampicin sensitive (RS-TB) tuberculosis. Methods We performed a secondary analysis of data from a cross sectional study that enrolled adult participants with bacteriologically confirmed pulmonary TB at a national tuberculosis treatment center in Uganda. Blood samples were tested for CD4 and CD8 cell counts, HIV serology and a full hemogram. Rifampicin sensitivity and the bacillary load grade were determined by Xpert MTB/RIF®. Fifty-five participants that had RR-TB (cases) were matched with 110 participants that had RS-TB (controls) for age, sex and HIV status in a ratio of 1:2 respectively. Sensitivity (Se), specificity (Sp), area under curve (AUC) analysis and determination of optimal cut-offs were performed using receiver operating characteristic curves. Results Cases differed from controls with respect to residence (p = 0.031), bacillary load grade (p < 0.010) and MCV (p = 0.021). The Se, Sp and AUC of the MCV (cut-off of > 74.6 femtolitres (fl)) were 88.9%, 34% and 0.607 (p = 0.021) respectively for RR-TB. Among HIV positive participants, the respective Se, Sp and AUC of the MCV for RR-TB (cut-off of > 72.5 fl) were 97.2%, 22.2% and 0.608 (p = 0.061). The respective Se, Sp and AUC of the CD4/CD8 ratio (cut-off of > 0.40) were 67.3%, 50.0% and 0.559 (p = 0.199) on the overall and 54.1%, 71.6% and 0.628 (p = 0.024) among the HIV positive participants for RR-TB. Conclusion The MCV had a high sensitivity but very low specificity for RR-TB. The CD4/CD8 ratio had a low sensitivity and specificity for RR-TB among HIV positive individuals. The utility of either test is low due to low diagnostic accuracy.