Abstract
Background
In the early stages of a novel pandemic, testing is simultaneously in high need but low supply, making efficient use of tests of paramount importance. One approach to improve the efficiency of tests is to mix samples from multiple individuals, only testing individuals when the pooled sample returns a positive.
Methods
I build on current models which assume patients’ sero-status is independent by allowing for correlation betweenconsecutive tests (e.g. if a family were all infected and were all tested together). In this model, I simulate 10,000 patients being tested in sequence, with population sero-prevalence ranging from 1% to 25%, using batch sizes from 3 to 10, and assuming the increased probability of consecutive infections ranged from 0% to 50%.
Results
I find that as the likelihood of consecutive infected patients increases, the efficiency of specimen pooling increases. As well, the optimal size of the batch increases in the presence of clustered sequences of infected patients.
Heat map indicating the manner in which the number of tests needed is reduced as population prevalence and correlation between cases changes. Red indicates that there is no reduction in the number of tests, and blue indicates a near 100% reduction in the number of tests, with intermediate colors indicating intermediate fractions.
Conclusion
This analysis indicates further improvements in specimen pooling efficiency can begained by taking advantage of the pattern of patient testing.
Disclosures
Jeffrey Rewley, PhD, MS, American Board of Internal Medicine (Employee)