Pooled testing for HIV viral load: simulation of reaction use and time to confirmation of failure, with parameters from the Brazilian public network
Year: 2026
Type: article
Abstract: Background: in HIV treatment programs with high viral suppression, most routine viral load tests fall below the failure cutoff. In Brazil, about 89% of results from people on treatment for six months or more are below 50 copies/mL, and the public network contracts around 1.1 million tests per year. Testing samples in pools saves reactions, but dilution, low-level viremia and residual RNA below 50 copies/mL may delay the detection of failure at the 200 cutoff or consume the savings. This note records the technique and estimates its potential, using Brazilian parameters as a reference case. Methods: simulation with the national viral load distribution (2022), detectable residual RNA in 25% of collections below 50 copies/mL in the main scenario, drawn between 1 and 49 copies/mL (in the sensitivity analysis, shares from 0% to 40% and ranges from 20 to 49 and from 40 to 49) and the detection, quantification and reproducibility described in the instructions for use of the network kits (cobas HIV-1 on the 5800 and 4800 systems). Retest rules were restricted to those the kit allows to be executed: retest only pools quantified at 20 cp/mL or more (rule Q20, main) or any detected pool (rule D). In the cross-sectional stage (400,000 collections), strategies were compared by subrange, with discordances in both directions. In the longitudinal stage, a cohort of 200,000 suppressed adults was followed for 24 months, with collections every six months, confirmation in a new collection 28 days after the report and the same 17,301 episodes in both arms. Variation comes from 40 replicates. Results: pools of 4 with rule Q20 used 39.7% fewer reactions per collection (empirical interval across replicates 39.5% to 40.0%) and 50.4% fewer during follow-up (50.3% to 50.5%), with a simulated difference close to zero in confirmation within 60 days (0.0 point; −0.3 to 0.2) and a median of 47 days until confirmation, against 42 for individual testing. In the residual RNA scenarios tested, rule Q20 saved 26% to 42% of reactions per collection and 35% to 53% during follow-up; rule D, from 38% to nothing. Just above the cutoff, confirmation within 60 days had wide intervals (rebounds of 200 to 249: −1.0 point; −6.7 to 4.2) and fell by about 3 points between 200 and 299 with greater imprecision. Quantitative deconvolution saved 56% of reactions, but detected only 35.0% of loads between 200 and 249, against 65.5% for individual testing. Pre-concentration of plasma showed no advantage: with rule D, it saved 6% to 7%; with a trigger equivalent to that of Q20, it tied with the ordinary pool. Conclusion: in this simulation, retesting only quantified pools kept savings in all the scenarios tested, from 26% to 53% of reactions, with a magnitude sensitive to the distribution of loads below 50, and confirmation of failure close to that of individual testing overall, with more uncertainty just above the cutoff. The cost reduction compared with current practice would be a fraction of these percentages, proportional to the eligible fraction and to the weight of the reaction in the price, and would depend on laboratory validation, on admission of the procedure and on payment linked to the number of reactions.
Author: Wistuba Melo da Cunha, Willian
Language: English
Subfield: Infectious Diseases
Field: Medicine
Domain: Health Sciences
Open Access status: green
OpenAlex ID: W7225620366
FWCI:
Cites:
Cited by:
Related to: