Proactive COVID-19 testing to mitigate spread
0 likes1,117 views
1. Carl Bergstrom discusses the role of proactive COVID-19 testing in mitigating the spread of the virus. He outlines four roles for testing: individual diagnosis, clearance, surveillance, and mitigation. 2. Bergstrom presents analytic models showing that regular testing can significantly reduce exposure days by identifying infectious individuals who would otherwise spread the virus without knowing they are infected. The models factor in testing frequency, false negative rates, and delays in getting results. 3. More complex stochastic models developed by Ryan McGee capture real-world factors like disease dynamics across social networks, individual heterogeneity, and varying test sensitivity over time. Simulations show that even with these complexities, regular testing as part of a test-trace
Proactive COVID-19 testing to mitigate spread
- 1. COVID-19 Proactive testing to mitigate spread Carl T. Bergstrom Department of Biology University of Washington IPAM Workshop on Mathematical Models in Understanding COVID-19 Photo: Carl Bergstrom
- 2. Analytic approximations Ted Bergstrom, UCSB Haoran Li, UCSB Stochastic network model Ryan McGee, UW with Ben Kerr, Omar Cornejo, Mark Tanaka, Julian Homberger, Hannah Williams, Alicia Zhou Photo: Carl Bergstrom
- 3. Disclosures I have a paid consulting relationship with Color Genomics, and I will be talking about some of that work today. I have no financial interest in the production or sales of COVID tests or treatments. I do have vested interests in going places, seeing people, working at the office, eating at restaurants, drinking at the bar, playing sports, hearing live music, sending my son to school, sending my daughter to college, visiting Europe, visiting anywhere else, and photographing birds that cant be bothered to visit Seattle. Photo: Carl Bergstrom
- 4. A vaccine is coming But it wont be a panaceaif it comes at all. Photo: Carl Bergstrom
- 5. We need something else Something nimble. Photo: Carl Bergstrom
- 6. This is a talk about testing. Photo: Carl Bergstrom
- 7. Four roles for Covid-19 testing Individual diagnosis, clearance, surveillance, and mitigation. 1. Individual diagnosis Symptomatic patients, for treatment and peace of mind. High sensitivity and specificity desired. CC-BY-2.0 Lisa Helfert
- 8. Four roles for Covid-19 testing Individual diagnosis, clearance, surveillance, and mitigation. 2. Clearance Verify that patients and practitioners are uninfected prior to performing a procedure. Travel? Other activities? High sensitivity needed. Specificity less essential. CC 59th Medical Wing
- 9. Four roles for Covid-19 testing Individual diagnosis, clearance, surveillance, and mitigation. 3. Surveillance Public health officials want to track prevalence and trajectory of pandemic. Can correct for lower sensitivity and (if common) lower specificity. Chen and Ngu 2020 ProPublica
- 10. Four roles for Covid-19 testing Individual diagnosis, clearance, surveillance, and mitigation. 4. Mitigation Find non-symptomatic cases and isolate / trace as a means of disease control. Volume is more important than sensitivity or specificity. Larremore et al. 2020 Test sensitivity is secondary to frequency and turnaround time for COVID-19 surveillance
- 11. Speed is critical in all cases. 1. Individual health. Patients dont want to wait. Treatments may depend on diagnosis. Helfert; 59th Med Wing; ProPublica; Larremore 3. Surveillance Heath officials need the latest information to make good decisions. 4. Mitigation Every days delay is an extra day an infected person is walking around spreading disease. 2. Clearance Tests should be point-of- care to minimize chance of becoming infectious after sampling.
- 12. The CDC guidelines state that: Testing of all students, faculty and staff for Covid-19 before allowing campus entry (entry testing) has not been systematically studied. It is unknown if entry testing in IHEs provides any additional reduction in person-to-person transmission of the virus beyond what would be expected with implementation of other infection preventive measures (e.g., social distancing, cloth face covering, hand washing, enhanced cleaning and disinfection). Therefore, CDC does not recommend entry testing of all returning students, faculty, and staff.
- 13. The aim of proactive (mitigation) testing: Reducing exposure-days Latent Infectious Time Test quarantine Delay d Infectious period C Results
- 14. C # infectious days n Testing periodicity q False negative rate d Delay for results By what fraction does testing and isolation reduce exposure days relative to no testing at all?
- 15. C # infectious days n Testing periodicity q False negative rate d Delay for results When testing periodicity is longer than infectious period, individuals will only be tested once while infectious. true positive false negativetested not tested Simplifies to Expected exposure days
- 16. When testing periodicity is shorter than the infectious period, you get multiple chances to catch an infection. Where and C # infectious days n Testing periodicity q False negative rate d Delay for results After a bit of algebra, you can still get a nice expression for the expected exposure days.
- 17. But some symptomatic individuals can self-isolate. We need to account for this. Time Latent Asymptomatic Exposure days u Latent Pre-sympt. Symptomatic, non-isolating Exposure days (1-u)v Latent Pre-sympt. Sympt. Self-isolating Exposure days (1-u)(1-v)
- 18. Compare infectious days with testing to infectious days without.
- 19. Compare infectious days with testing to infectious days without. Asymptomatic or dont isolate Self-isolate
- 20. Compare infectious days with testing to infectious days without. Asymptomatic or dont isolate Self-isolate
- 21. Results
- 22. The real world is not so simple Disease dynamics Social network structure Heterogeneity of disease trajectory, infectivity, etc. Varying test sensitivity over time Non-compliance with testing and isolation Photo: Carl Bergstrom
- 23. Ryan McGees SEIRS+ framework Open source python-based stochastic SEIRS model with multi-level network structure, network interventions (targeted testing, social distance, contact tracing, isolation, etc.) https://github.com/ryansmcgee/seirsplus
- 25. Testing, tracing, and isolation https://github.com/ryansmcgee/seirsplus
- 28. Test sensitivity Kucirka et al. 2020 Ann. Int. Med. https://github.com/ryansmcgee/seirsplus
- 29. Use case: workplace testing https://www.color.com/covid-19-outbreak-model Setting: workplaces of size 50-1000. Intervention: self-administered workplace testing via nasal swab. Network: single-layer FARZ network with 40% global transmission Outcome measure: size of epidemic resulting from one introduction
- 33. Testing helps a lot https://www.color.com/covid-19-outbreak-model Total epidemic size
- 34. Outbreak sizes https://www.color.com/covid-19-outbreak-model Mean fraction infected Fraction of introductions infecting >5%
- 35. Speed is of the essence https://www.color.com/covid-19-outbreak-model Mean fraction infected Fraction >5% Total epidemic size
- 36. Use case: community test-trace-isolate Setting: Community of 50,000 Intervention: test-trace-isolate Network: multilayer FARZ with household, school, workplace structure. Outcome measure: change in effective R value
- 37. Test and isolate only
- 38. Test, trace, and isolate Households of positive tests are isolated. Contacts traced in two days, and isolated along with household.
- 39. Comparing with the analytic approximation
- 40. Comparing with the analytic approximation
- 41. Pooled testing stretches capacity.
- 42. Pooled testing stretches capacity.
- 43. Pooled testing stretches capacity.
- 44. Take-home messages Proactive testing can help control the epidemic but the speed of turnaround is essential. Simple models give decent approximations and allow quick exploration of speed / sensitivity / volume / cost tradeoffs. Network structure and heterogeneity matterespecially for rapid spread. Photo: Carl Bergstrom