Low Testing Rates Limit the Ability of Genomic Surveillance Programs to Monitor SARS-CoV-2 Variants: A Mathematical Modelling Study

Background: Genomic surveillance is essential for monitoring the emergence and spread of SARS-CoV-2 variants. SARS-CoV-2 diagnostic testing is the starting point for SARS-CoV-2 genomic sequencing. However, testing rates in many low- and middle-income countries (LMICs) are low (mean = 27 tests/100,000 people/day) and global testing rates are falling in the post-crisis phase of the pandemic, leading to spatiotemporal biases in sample collection. Various public health agencies and academic groups have produced recommendations on sample sizes and sequencing strategies for effective genomic surveillance. However, these recommendations assume very high volumes of diagnostic testing that are currently well beyond reach in most LMICs.Methods: To investigate how testing rates, sequencing strategies and the degree of spatiotemporal bias in sample collection impact variant detection and monitoring outcomes, we used an individual-based model to simulate COVID-19 epidemics in a prototypical LMIC. Within the model, we simulated a range of testing rates, accounted for likely testing demand and applied various genomic surveillance strategies, including sentinel surveillance.Findings: Diagnostic testing rates play a substantially larger role in monitoring the prevalence and emergence of new variants than the proportion of samples sequenced. To enable timely detection and monitoring of emerging, programs should achieve average testing rates of at least 100 tests/100,000 people/day and sequence 5-10% of test-positive specimens, which may be accomplished through sentinel or other routine surveillance systems. Under realistic assumptions, this averages to ~10 samples for sequencing/1,000,000 people/week.Interpretation: For countries where testing capacities are low and sample collection is spatiotemporally biased, surveillance programs should prioritize investments in wider access to diagnostic testing to enable more representative sampling, ahead of simply increasing quantities of sequenced samples. Funding Information: This research was supported by the European Research Council, the Rockefeller Foundation, and the Governments of Germany, Canada, UK, Australia, Norway, Saudi Arabia, Kuwait, Netherlands and Portugal.Declaration of Interests: A.T., E.H., S.C., B.R. and B.E.N. declare that they are employed by FIND, the global alliance for diagnostics. The remaining authors have nothing to declare.