1. Residual Risk and Waste in Donated Blood with Pooled Nucleic Acid Testing Hrayer Aprahamian, Dr. Ebru Bish, Dr. Douglas Bish Virginia Polytechnic Institute and State University, Dept. of Industrial and Systems Engineering, Blacksburg, VA Research supported by the National Science Foundation October 8, 2015

2. Motivation Blood products are vital healthcare commodities. Transfusion-transmitted infections (TTI) include: HIV, HBV, HCV, WNV and Babesiosis (among others). Even with advances in testing technology, the risk of transmitting an infection through blood transfusion remains. Effectively managing the limited resources to improve the safety of blood is crucial. Accurately measuring this risk is of utmost importance to aid with strategic decision-making. 2

3. The Challenges Screening tests are not perfectly reliable. FDA requires or recommends a set of infections for which the blood needs to be tested in the US, but does not specify a testing strategy. Nucleic Acid Tests (NAT) are more sensitive but are considerably more expensive and time-consuming: Individually testing using NAT might not be a feasible option. 3

4. The Challenges Pooled testing is often used to reduce testing costs: Dilution effect. Increases waste figures. In practice, failed samples are re-tested. Numerous re-testing schemes are available. 4

5. The Challenges Residual risk alone does not provide the complete picture. 5 Residual risk Testing cost Expected waste Testing strategy

6. The Challenges Individual variability in viral growth. In sample test variability (i.e., testing the same sample multiple times may yield different outcomes). Infectivity of the blood sample: Low viral loads in blood may not cause infection. 6

7. Research Objective 7

8. Literature Review Transfusion: Mostly focus on blood screening and impose certain assumptions: E.g., Busch 2000, Weusten et al. 2002, Jackson et al. 2003, Busch et al. 2005, Glynn et al. 2005, Weusten et al. 2011. Operations Researchers and Statisticians: Approach the problem from the perspective of pooled testing: E.g, Dorfman 1943, Arnold 1977, Wein and Zenios 1996, Kim et al. 2007, Habtesllassie et al. 2015. 8

9. Viral Model 9

10. 10 Infectivity

11. Testing Error 11

12. Re-test Schemes 12 m

13. Re-test Schemes 13 m

14. Re-test Schemes 14

15. Re-test Schemes Decision rule is crucial for determining the outcome of the re-test: We adopt the Believe the positive decision rule for all re-test schemes (except array based). In the array re-test scheme a sample is rejected if at least one of the columns AND one of the rows fail corresponding to that node fails. 15

16. Re-test Schemes 16 Re-test #1Re-test #2

17. Events 17

18. Residual Risk (RR) 18

19. 19 Expected number of tests only depends on the outcome of the initial pool.

20. Analysis – Residual Risk Lemma 1.

21. Analysis – Expected Tests Lemma 2.

22. Analysis – Expected Waste Lemma 3.

23. Case Study We perform a case study using data available on South Africa. Some major findings are: Performing re-tests increases both the risk and the expected number of test. The benefits of re-tests is to reduce the expected waste. Adding a single re-test drastically reduces the waste figures. Array-based re-testing schemes outperforms other schemes when: Testing costs are high. Prevalence rates are high 23

24. Cost-effectiveness 24 Best solution Current practice

25. Cost-effectiveness 25

26. Conclusions Determining the “best” testing strategy is not a trivial matter. The expressions derived provides an accurate approach to weight in the trade-offs being incurred to determine a suitable testing strategy. 26

27. Reference and Contact Info. Aprahamian, H., Bish, D., and Bish, E. “Residual Risk and Waste in Donated Blood with Pooled Nucleic Acid Testing ”, submitted for publication. Email: ahrayer@vt.edu 27

28. Thank you 28