Accuracy of WHO Immunological Criteria in Identifying Virological Failure among HIV-infected Adults on First line Antiretroviral therapy in Mwanza, North-western Tanzania.

Optimal HIV treatment monitoring remains a big challenge in resource limited settings. Guidelines recommend the use of clinical and immunological criteria in resource limited settings due to unavailability of viral load monitoring; however their utility is questionable. This study aimed at assessing the accuracy of immunological criteria in detecting treatment failure among HIV infected Tanzanian adults receiving first line ART. A clinic based cross sectional study was conducted between February and July 2011 at Bugando Medical centre (BMC) HIV care and treatment clinic (CTC) involving HIV infected patients aged 18 years and above, receiving first line ART; followed up for at least 1 year. Viral load was tested for every enrolled patient. Standard WHO criteria were used to define immunological failure. Virological failure was defined as one viral load measurement of >5000 copies/ml and was used as a gold standard. A 2 × 2 table was used to assess the accuracy of immunological criteria in detecting treatment failure. A total of 274 HIV-infected adults were enrolled into the study. Out of these, 65.7% were females, the median age was 39 years (IQR 33-45), the median BMI 21.9 kg/m(2) (IQR 19.7-24.0). Out of the 274 study participants 156 (56.9%) had immunological failure. Only 60 of the study participants (21.9%) had viral load >5000. Only 42 patients (70%) were found to have both immunological failure and virological failure. The sensitivity of immunological criteria in detecting treatment failure was 70%, specificity 46.7%, positive predictive and negative predictive values of 26.9 and 84.7% respectively. WHO immunological criteria have low sensitivity and positive predictive value for detecting treatment failure. Relying on CD4 counts for treatment monitoring would therefore lead to misclassifications of treatment failure that could result into unnecessary or delayed switch to second line ART. Access to viral load monitoring is important to avoid these misclassifications