CROI 2018 Abstract eBook

Abstract eBook

Poster Abstracts

Twenty-nine GeneXpert instrumentation errors occurred, yielding an error rate of 1.8%. Conclusion: Findings indicate that at both the WHO cutoff of 1000 copies/mL and the Botswana cutoff of 400 copies/mL, the Xpert HIV-1 VL performs well compared to the CAP/CTM HIV-1 v2 in a setting of intended use. These findings suggest that the assay is suitable to complement conventional platforms, particularly for target populations in resource-limited settings where VL testing is lacking.

1143 THE ECONOMIES OF SCALE OF TEST AND TREAT: A LONGITUDINAL COSTING STUDY IN SWAZILAND Shaukat Khan 1 , Cebele Wong 1 , Pascal Geldsetzer 2 , Gerda Asmus 3 , Thomas How 1 , Charlotte Lejeune 1 , Velephi Okello 4 , Till Bärnighausen 3 1 Clinton Health Access Initiative, Mbabane, Swaziland, 2 Harvard University, Cambridge, MA, USA, 3 Heidelberg University, Heidelberg, Germany, 4 Ministry of Health, Mbabane, Swaziland Background: Swaziland has one of the highest adult HIV prevalence rates worldwide, 27%; with approximately 200,000 people 15 years and older living with HIV. Swaziland and many other countries in sub-Saharan Africa have adopted universal test-and-treat (UTT) policies. As the number of people needing and receiving ART in Swaziland is rapidly increasing, it is critical to understand the cost of UTT scale-up and its implications for the health system. The study presented here is the first empirical longitudinal costing study of annual ART patient costs under a public-sector UTT policy with routine viral load monitoring. Methods: We collected comprehensive monthly facility-level data on ART patient costs from 14 facilities implementing UTT as part of a large-scale randomized stepped-wedge health systems trial (September 2014-December 2016). In addition to a comprehensive time-and-motion study, we used extracted cost data from facility budgets, expenditure reports, and patient records. Items included in this “bottom-up costing” included direct personnel, medications, laboratory services including viral load, and treatment for opportunistic infections (OI). We express all costs per patient per year (PPPY). Costs were converted from local currency to U.S. Dollars using annualized exchange rates. Results: Total ART costs PPPY were $214 (95% CI: 201-226). ARVs costs accounted for the largest proportion at $102 (95% CI: 101-103), followed by personnel $77 (95% CI: 67-88), laboratory services (including viral load) $31 (95% CI: 27-35) and OI costs $4 (95% CI: 3-4). In the descriptive data (Figure 1) and in multilevel regression analysis controlling for time fixed effects and facility random effects, we identified strong economies of scale in the relationship between costs PPPY and facility size (measured in the number of visits per month). As facility size increases, costs initially decrease rapidly (from about $800 PPPY) and then plateau (below $200 PPPY) approximately at 250 patient visits. Conclusion: Swaziland’s public-sector ART program displays strong economies of scale under UTT, with far less efficiency achieved in clinics reaching less than 250 patient visits per month. In the context of scaling up UTT, increases in patient volumes are efficient trajectories but where not feasible (such as rural and remote populations), alternative delivery models could provide efficiency gains; including community health worker delivered ART.

Poster Abstracts

1142 EXPANDING VIRAL LOAD TESTING ACCESS THROUGH EVALUATION OF THE GENEXPERT IN BOTSWANA Rachel S. Beard 1 , Anikie Mathoma 2 , Mackenzie Hurlston 1 , Naledi Nyepetsi 3 , Christopher Serumola 2 , Joyce Basotli 2 , Ritah Moatlhodi 3 , Unami Mathebula 2 , Otsile Meswele 4 , Charity Ralegoreng 4 , Tefera Agizew 2 , Rosanna Boyd 2 , Madisa Mine 5 , Ishmael Kasvosve 3 1 CDC, Atlanta, GA, USA, 2 CDC Botswana, Gaborone, Botswana, 3 University of Botswana, Gaborone, Botswana, 4 African Society for Laboratory Medicine, Gaborone, Botswana, 5 Botswana Ministry of Health, Gaborone, Botswana Background: With the UNAIDS target to reach 90% HIV viral suppression of those on treatment by 2020, access to viral load (VL) testing is essential. However, VL testing for monitoring of treatment efficacy remains limited and confined to centralized laboratories with demanding infrastructural and technical requirements. Large field evaluations for point of care (POC) and near POC VL instrumentation and assays are lacking. We assessed the performance and operational characteristics of the Xpert HIV-1 VL assay and the GeneXpert near (POC) platform, in clinics with limited laboratory infrastructure in Botswana. Methods: In a multi-site cross-sectional study, people living with HIV, including those currently on antiretroviral therapy (ART) and ART naïve patients, were enrolled beginning in May 2016 from four HIV clinics, 2 in Francistown and 2 in Gaborone. Plasma samples from consenting patients were prepared and tested on Xpert HIV-1 VL and a reference assay, Roche COBAS AmpliPrep/ COBAS TaqMan HIV-1 v2.0 (CAP/CTM HIV-1 v2), according to manufacturers’ recommendations. Correlation was assessed via linear regression. Sensitivity and specificity were determined at the World Health Organization (WHO) recommended clinical cut-off of 1000 copies/mL and the Botswana National Guidelines’ cut-off of 400 copies/mL. Results: Of the 1600 patients enrolled, 1597 patients’ plasma were tested on both the Xpert HIV-1 VL and the reference assays. Of the plasma samples tested, 583 (36.5%) are from patients on ART. Gender information was available for 1585 patients of which 1379 (87.0%) were female. Sensitivity and specificity of Xpert HIV-1 VL from the 1597 tested was 98.2% and 98.7% at 1000 copies/mL and 97.7% and 98.6% at 400 copies/mL, respectively. Linear regression analysis (Figure 1) demonstrated good correlation between Xpert HIV-1 VL and CAP/CTM HIV-1 v2 ( R2 =0.897). The upward and downward misclassification was 1.34% and 1.81% at 1000 cps/mL and 1.42% and 2.32% at 400 cps/mL, respectively.

CROI 2018 442