(P104) HOW COMPLEX IS POOLED TESTING FOR STI DETECTION? A NEW METRIC TO EVALUATE ITS IMPLEMENTATION

Background:
Pooled testing has long been used to test large-volume clinical specimens for sexually transmitted infections. The reason is because fewer tests are needed on average by using pooled testing rather than when testing each specimen separately. There are many strategies to implement pooled testing. The simplest is to pool specimens and test this pool for a pathogen. If the pool is negative, the corresponding individuals represented within it are declared negative. If the pool is positive, portions of the same specimens are retested separately to determine a positive/negative outcome for each individual. Other strategies exist to implement pooled testing. And, many of these strategies are much more efficient (i.e., result in fewer tests). However, more efficiency can come with more complexity when implementing a strategy.

Methods:
We developed a new statistical metric to represent complexity for pooled testing. This metric is found for the previously described strategy (formally known as “Dorfman testing”) and also other strategies, such as matrix pooling. We constructed a web-based app (www.chrisbilder.com/app) to compute this metric that provides a user-friendly interface for those without statistical experience. The R statistical software package provides the underlying calculation environment.

Results:
We applied the metric to a simulated setting motivated by how testing for Chlamydia trachomatis (CT) and Neisseria gonorrhoeae (NG) is performed in Iowa. We showed that modest gains of efficiency (approximately 5%) can be obtained by using matrix pooling rather than the current practice of Dorfman testing. However, matrix pooling is 52% (60%) more complex than Dorfman testing for CT (GC) detection.

Conclusion:
Our proposed metric gives laboratories a new tool to examine pooled testing prior to being put into practice. This metric provides an analytical way to evaluate different strategies so that one can determine whether efficiency gains from a strategy are worth an increase in complexity.