There are a few peer-reviewed studies published in major medical journals that break things out that way, Fuji. The problem with that as a meaningful source of data on the subject is that they are inherently a collection of anecdotes. EVERY death is unique, obviously, because so many factors come together in a fatal reaction-- but the larger issue is that anaphylaxis deaths are NOT reportable to any agency. Not even internationally, at least that I am aware. It is possible that NHS keeps such records in the UK, but even if they did, one need only examine the peer reviewed studies that find that at least half of anaphylaxing patients are not recognized and treated properly by EMT/EMS or by emergency room physicians either one. It's unlikely that they'll be coded with anaphylaxis as COD if a physician didn't recognize an allergic reaction in-progress while they were trying to keep the patient alive.
KWIM?
The numbers are low-- but nobody is really sure just how low. This is a bit like looking at hospital errors, though-- there is a huge amount of inertia in having solid numbers there, because they certainly aren't going to make emergency responders and ER's look great.