You may have heard about the recent enterohaemorrhagic E. coli (EHEC) outbreak in Germany.
What questions would a statistician ask about EHEC analysis ?
I'm thinking of Q+As between reporters / public officials ↔ non-experts, say teachers and engineers with a Diplom / Master's degree but at most a smattering of statistics.

(Is a picture, a map of EHEC land showing various strains of EHEC and the coverage of various tests, possible ?)

Monday 20 June: I thought that the EHEC outbreak would be an area where statistics really matters in the world at large: what's the evidence for various causes, how can these be communicated to the public ? So, starting a bounty.

  • 1
    $\begingroup$ I'm not sure if it should be community wiki as I think that votes on good answers should count towards rep, and there's a chance you'll get better answers if such is the case. $\endgroup$ – Jeromy Anglim Jun 14 '11 at 11:15

If you have no a priori scientific questions you want/need answered - I would look at the spatial distribution of cases. In particular, I would estimate a sort of spatial "intensity" function-- that is some function $f( {\bf s} ) = E({\rm prevalence} | {\rm \ you \ are \ in \ location \ } {\bf s} )$. So that you don't overfit the data, some smoothing would be appropriate, which would effectively "bin" nearby locations together. The simplest approach would be using a kernel density estimation function (e.g. the kde2d function in R, although some pre-processing of the data may be necessary). This would help identify "hot spots" in the spatial distribution and could lead an investigator to identify environmental reservoirs of E. Coli.

Depending on what other data you have, more sophisticated spatial models that adjust for known confounders could be used.

  • $\begingroup$ Diagnosing EHEC variants and their possible sources seems difficult, iffy. As you know, German officials first said "it's cucumbers from Spain", then weeks later "it's definitely sprouts" (seeds, water ... ?) (I realize that many people want an instant answer, and are served one.) With clear symptoms, a map may lead to a cause, as in the famous map of the cholera outbreak in London in 1854. But EHEC outbreak map ? $\endgroup$ – denis Jun 22 '11 at 16:13

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.