Is there a way to categorise harm (eg from woo remedies)?

How do we categorise harm? 

(i) If I eat some bits of a yew or some iffy fungus I'll be very unwell - this is harm arising because the substance itself is toxic (clearly dose has some impact).

(ii) If I drink a lot of grapefruit or orange juice while taking certain statin drugs I'll reduce the rate at which my body clears the statin from my system, this might cause a problematic increase in the drug - this is harm arising from the grapefruit interacting with the enzyme that's meant to be clearing statins from the body (interactions).

(iii) If I buy some dodgy herbal pills from the internet they might contain prescription-only medicines that I don't know about. The real medicine could have been withdrawn from sale, or could interact with other prescribed meds that I might be taking or something else - this is harm arising from insufficient information and also a bit of (i) and (ii).

(iv) If I have a potentially serious health problem but choose to take treatment from an unconventional healer then by delaying getting appropriate treatment I may become very ill - this is harm arising from failure to act to preserve health.

There are probably other nuanced versions of these - I'm wondering if there's a recognised typology of harm, in the same way that you can have a Type I or Type II error in statistics.

If not, can we make some up ourselves?

Edit 31 August 2014
Just read an interesting post from Edzard Ernst looking more closely at the link between cardiac patients who are taking herbal remedies and their adherence to their prescribed medication. It looks like there may be a link (perhaps not surprising, though possibly not studied in depth before) and it seems that ther'e's a correlation between taking herbals and not taking prescribed medication appropriately. This could be dangerous and relates to (iv) in my imaginary taxonomy above.

A hitherto unknown risk of herbal medicine usage (31 August 2014) Edzard Ernst's blog

----

A project I work on (CHI+MED - making medical devices safer) looks at many aspects of medical safety, including human factors and systems thinking in handling medical errors. Specifically this involves looking at ways of designing into the system or device ways of making unavoidable user error more noticeable so that people can recover from them.

The older 'blame culture' that's been prevalent in many healthcare systems has taken the view that error is because someone's done their job wrongly and the response has been to retrain them. If you've ever poured orange juice in your tea or forgot your umbrella you can see immediately that this isn't a helpful view to take. Human error is pervasive (hence inevitable) and only rarely will training (or worse, sacking and getting in new people) fix it. Much better to learn from error and bolster systems to protect against it.

To a certain extent Google does this everytime you mistype something and it says "did you mean?" and spellcheckers do something similar for Word documents. In both cases the system has a design function that acknowledges the possibility of mistyping and offers an alternative or solution. Similarly most keyboards have a delete key to let you undo and even pencils have an eraser on the end of them.

We've found a really nice way of talking about error that doesn't involve blame - the dumb things we do everyday tend to be quite funny and no-one really seems to mind poking fun at themselves for doing something silly. And lo and behold, the cognitive processes involved in making these everyday errors are pretty much identical to those often involved in medical error - so we can learn from them too - have a look at the #errordiary hashtag and the Errordiary website which explains more.