Abstract
Background: The first pillar of the UNAIDS 90-90-90 goal seeks to accurately identify persons living with HIV (PLHIV), a
process that is predicated on facilities having the necessary HIV tests available to perform the task. In many rural settings,
the identification of PLHIV is accomplished through a two-step process involving the sequential use of 2 separate rapid
diagnostic tests (RDTs). Inadequate inventory of either test has ramifications for the success of HIV-related programs. The
purpose of this study was to evaluate the inventory levels of HIV RDT kits at specific healthcare facilities in Zambézia
province, Mozambique.
Methods: Using facility-level pharmacy stock surveillance data from October 2015 through September 2016, we assessed
the inventory levels of HIV RDTs at 75 health facilities in 8 districts within Zambézia province, Mozambique. Using
programmatically established categories (good, sufficient, threatened, or stockout), defined in conjunction with the
provincial health authorities, descriptive statistics were performed to determine inventory control of HIV RDTs at the district
and health facility levels. Monthly proportions of adequate (good + sufficient) inventory were calculated for each district to
identify inventory trends over the evaluation period. To assess whether the proportion of inadequate stocks differed between
RDT, a mixed-effects logistic regression was conducted, with inadequate inventory status as the outcome of interest.
Results: When viewed as a whole, the inventory of each test kit was reported as being at adequate levels more than 89% of
the time across the 75 facilities. However, disaggregated analysis revealed significant variability in the inventory levels of HIV
RDTs at the district level. Specifically, the districts of Inhassunge, Namacurra, and Pebane reported inadequate inventory
levels (threatened + stockout), of one or both test kits, for more than 10% of the study period. In addition, a disparity between
inventory levels of each test kit was identified, with the odds of reporting inadequate inventory levels of the confirmatory test
(Uni-Gold™) being approximately 1.8-fold greater than the initial test (Determine™) (odds ratio: 1.82, 95% CI: 1.40-2.38).
Conclusion: As Test and Treat programs evolve, a significant emphasis should be placed on the “test” component of the
strategy, beginning with assurances that health facilities have the adequate inventory of RDT necessary to meet the needs
of their community. As national policy-makers rely predominantly on data from the upstream arm of the supply chain, it
is unlikely the disparity between inventory levels of HIV RDTs identified at individual districts and specific health facilities
would have been recognized. Moving forward, our findings point to a need for (1) renewed efforts reinforcing appropriate
downstream forecasting of essential medicines and diagnostic tests in general and for Uni-Gold™ test kits specifically, and (2)
simple metrics that may be routinely collected at all health facilities and which may then easily and quickly flow upstream so
that policy-makers may optimally allocate resources.