Research Article: Economic and public health impact of decentralized HIV viral load testing: A modelling study in Kenya

Date Published: February 27, 2019

Publisher: Public Library of Science

Author(s): M. de Necker, J. C. de Beer, M. P. Stander, C. D. Connell, D. Mwai, David P Wilson.


Kenya has the world’s 4th largest HIV burden. Various strategies to control the epidemic have been implemented, including the implementation of viral load (VL) testing to monitor HIV patients on ARVs. Like many resource limited settings, Kenya’s healthcare system faces serious challenges in effectively providing quality health services to its population. Increased investments to strengthen the country’s capacity to diagnose, monitor and treat diseases, particularly HIV and TB, continue to be made but are still inadequate in the face of global health goals like the UNAIDS 90:90:90 which require scaling up of VL tests amid existing constraints. In Kenya, there is an increase in the demand for VL tests amidst these existing constraints. The GeneXpert system is a diagnostic point-of-care technology that can quantify, amongst others, HIV VL. Currently, GeneXpert technology is widely distributed in Kenya for testing of tuberculosis. This study aimed to determine the economic and public health impact of incorporating VL test modules on the existing GeneXpert infrastructure. Markov models were constructed for different populations (non-pregnant adults, pregnant women and children). The scenarios analysed were 100% centralized VL testing compared to 50% GeneXpert plus 50% centralized VL testing, with time horizons of 5 years for the adult and child populations, and 31 months for the pregnant population. Incremental effectiveness was measured in terms of the number of HIV transmissions or opportunistic infections avoided when implementing the GeneXpert scenario compared to a 100% centralized scenario. The model indicated that, for all three populations combined, the GeneXpert scenario resulted in 117 less HIV transmissions and 393 less opportunistic infections. The cost decreased by $21,978,755 for the non-pregnant and pregnant adults and $22,808,533 for non-pregnant adults, pregnant adults and children. The model showed that GeneXpert would cost less and be more effective in terms of total cost per HIV transmission avoided and the total cost per opportunistic infection avoided, except for the pregnant population, when considered separately.

Partial Text

Kenya exhibits one of the worst epidemics of HIV and AIDS in the world [1], with approximately 1.6 million Kenyans living with HIV and approximately 840 000 children orphaned due to the disease [2].

Markov models were constructed in Microsoft Excel with each model corresponding to one of three risk populations, namely non-pregnant adults (15 years and older), pregnant women and children (0 to 14 years). In this cohort model, two scenarios were compared for each of the three population risk pools. The model compared a scenario where 100% of VL testing is performed in centralized laboratories to a scenario where 50% of VL testing is done with GeneXpert plus 50% of VL testing is done in centralized laboratories.

Incremental effectiveness was measured in terms of the number of HIV transmissions or opportunistic infections avoided when implementing the GeneXpert scenario compared to a 100% centralized scenario. Incremental costs were calculated as the difference between the GeneXpert scenario and the 100% centralized scenario. Cost-effectiveness was calculated by dividing the incremental cost by the incremental effectiveness.

The premise of this study was that, while VL monitoring is the preferred strategy for PLHIV, it is also the most expensive option. Given the sunk cost of capital investment of GeneXpert infrastructure in Africa, we therefore questioned whether the adoption of VL monitoring on the back of this existing infrastructure would be a cost-effective intervention strategy. We answered this question by developing an economic model that mimics the natural history of three risk pools of PLHIV and estimating the economic cost of the intervention on the one hand and the public health impact it achieves on the other hand. To our knowledge, this study is unique in that we could not find any evidence of similar studies that benefit from an existing decentralized POC laboratory infrastructure or that stratifies risk pools into three different cohorts. From a policy perspective, we furthermore believe that the results would be relevant to stakeholders who consider adopting the new WHO guidelines related to monitoring of PLHIV.

For non-pregnant adults and children, introducing GeneXpert VL testing is a cost-effective way of implementing the WHO guidelines concerning monitoring of treatment failure and can be used to track progress toward the third 90-90-90 target.