Introduction: Early Infant Diagnosis (EID) targeting HIV exposed infants, happens over a period of time (6 weeks up to 18months), and employs different testing technologies at different testing time points. Being a process that is implemented over time and employing different testing technologies, a testing algorithm was designed to ensure reliable final diagnosis. This study was set out to estimate the level of adherence to the testing algorithm.

Methods: Retrospective EID data was collected from 24 health facilities that covered the entire tier of the health system, from 4 geographic regions of Uganda. From each study site, all exposed infants that were tested in 2012 were tracked from the 1st molecular test beginning at 6 weeks to the final rapid test at 18 months.

Results: From the 24 study sites, 4221 exposed infants were tested with 1st molecular test in 2012. Out of these, 3888 (92.1%) were HIV negative and 333 (7.9%) were HIV positive. Of the negative only 1543 (39.7%) did a 2nd molecular test representing a loss of 60.3%. A total of 14 babies converted to HIV with the 2nd molecular test, resulting in a total of 347 positive babies. Of the total positive babies, 212 (61.1%) initiated ART, representing a loss of 38.9%. Of the total who were screened with the 1st molecular test, 1130 (26.8%) were tested with the final rapid test at 18 months, representing a loss of 73.2%.

Discussion: The results indicate that, despite operational improvements, very poor adherence to EID testing algorithm was observed. This poor adherence may have an impact on the ultimate objectives of the whole program. Factors explaining high LTFU include; results long turnaround time (TAT), lack of patient follow-up mechanism, and poor patient counseling. There is therefore an urgent need to rethink the implementation of the EID program, not only in Uganda but also in other resourcelimited countries. Mechanisms of patient follow-up and linkage to care should be integrated into the testing process.