Read on Twitter
ZeroCOVID discussions have made contact-tracing-as-an-intervention a hot topic again. In particular, a contention frequently heard is that if we "get cases low" and then "build up the tracing system" test+trace can control cases.
A short thread of considerations and caveats.1/9
A short thread of considerations and caveats.1/9
1] Although this may come as a surprise, we're 1 year in and still no jurisdiction regularly shares data on the fraction of positive cases reached ahead of time by an out-of-household contact tracer.
This is the data that would be step 1 in an empirical case for this strategy.2/
This is the data that would be step 1 in an empirical case for this strategy.2/
For various other interventions (masks, closing restaurants, etc), determining effectiveness really is very difficult when effects are not large.
For contact tracing, *it should be possible to build a simple quantitative case for effectiveness* with a clear analysis of data. 3/9
For contact tracing, *it should be possible to build a simple quantitative case for effectiveness* with a clear analysis of data. 3/9
2] A frequent argument that tracing can make a big difference once cases are few is that places like Japan
+SK did well and have tracing. But there is a big logical fallacy here: tracing is just one thing these countries are doing, and they are not all doing it the same way. 4/9
+SK did well and have tracing. But there is a big logical fallacy here: tracing is just one thing these countries are doing, and they are not all doing it the same way. 4/9
Indeed
*)they aren't sharing data confirming high tracing effectiveness,
*)data released earlier as part of an age study in SK undercut the story of a very large effect
*)these countries fared better than others even after leaving suppression levels.5/9 https://twitter.com/WesPegden/status/1317970203167391744
*)they aren't sharing data confirming high tracing effectiveness,
*)data released earlier as part of an age study in SK undercut the story of a very large effect
*)these countries fared better than others even after leaving suppression levels.5/9 https://twitter.com/WesPegden/status/1317970203167391744
3] Finally, it is tempting to think that because retrospective tracing can reconstruct transmission chains (e.g., NZ was able to reconstruct a lot of past transmission chains in outbreaks) that with enough resources and few enough cases, tracing could prevent most transmission.6/
But this is a flawed conclusion. Indeed, reconstructing past transmission chains and preventing future ones are different goals, and, in particular, the first is made *easier* by the short serial interval of COVID, while the latter is made *much harder*. 7/9
Indeed, preventing most onward transmission primarily through tracing as part of a "zero COVID" strategy might be extremely difficult for precisely the same reason that careful resource-intensive investigations (e.g., in NZ) can often identify who-infected-who after the fact. 8/9
Certainly, tracing can prevent transmission, and pared with real support measures for isolation it makes sense that it can be an effective tactic.
Believing that getting cases low can produce a phase transition where TTI keeps us at ZeroCOVID does not seem supported by evidence.
Believing that getting cases low can produce a phase transition where TTI keeps us at ZeroCOVID does not seem supported by evidence.
