It is not possible to trace contacts that occur from a unit after that unit has been detected

Category:bug report
Priority:not yet set
Status:ready to commit

Infected units generate contacts which can result in the spread of disease. NAADSM records these contacts, so that they might be traced later. Tracing occurs when a truly infected unit is detected (see issue 2402 for a discussion of tracing from false-positive detections). All simulated trace investigations are initiated on the day after the infected unit is detected, although there may be a simulated delay in completing those traces. Tracing is triggered only once by the initial detection of an infected unit, and no further tracing from infected units ever occurs.

Although direct contact with an infected unit stops when that unit is detected (automatic quarantine), indirect contacts can continue. There is no way, then, to trace indirect contacts that occur after detection.

Is this behavior what we expect? Some discussion from the Development Team would be helpful, and we can then prepare appropriate documentation.





Just copying some notes I wrote via email into the discussion thread here...

(Describing the problem)

Suppose we have an infected unit, A.

On day 1, there is an indirect contact from A to B.
On day 2, disease is detected at unit A. A is quarantined. Trace forward/out discovers that there has recently been an indirect contact from A to B. Unit B is destroyed.
On day 3, because quarantine doesn’t stop indirect contact, there is an indirect contact from A to C.

Now, what happens to C?

Should C be destroyed? If the trace forward/out that occurred on day 2 is still “in effect”, in some sense, then C should be destroyed, just as B was.

But in the model as it is now, nothing happens to C. Tracing is an event that occurs on a specific day, and the trace looks for contacts that occurred in the past. The trace does not remain “in effect”, to catch future contacts involving the same unit.

I think this situation reveals there is some absurdity in combining the following two assumptions:
- Quarantine does not affect indirect contacts.
- It is possible to have a policy of pre-emptively destroying indirect contacts.

Put these two assumptions together and you get a somewhat bizarre mental image of quarantining a farm, but allowing the employees of that farm to continue driving trucks around, visiting other farms — even though every time they drive that truck onto another property, they instantly sign a death warrant for every animal there.

I think a decent temporary answer is just for people to be aware that any strategy that includes examining, testing, or pre-emptively destroying traced indirect contacts should also include deliberate reduction of indirect contacts (through movement controls). It just seems to make sense — if the authorities have some obligation to chase down indirect contacts and do something (visit them, test them, pre-emptively destroy them), then they will not want indirect contacts to continue unimpeded.

However, I generally don’t see people reduce indirect contact to zero in their simulations — they keep a small amount. In the QUADS Ireland simulations, it was assumed that a standstill order would achieve only a 68.5% reduction in indirect contacts inside zones. Is that to cover some types of contact that are unavoidable (maybe vets, feed?) or to reflect people disobeying the standstill order? I would guess that in the former case, you would not act on continuing indirect contacts from an already-detected farm, but in the latter case, you would.


Status:active» ready to commit