Monday, October 08, 2012

Risks, numbers and criminal responsibility

In R v Mabior, 2012 SCC 47 (5 October 2012) the Supreme Court of Canada playfully leaves us to calculate what a "realistic possibility" means in terms of probabilities, in the context of transmission of HIV virus by vaginal sexual intercourse.

Courts are inclined to put things in words, not numbers. This case explains that the phrase "significant risk of serious bodily harm" in this context means "a realistic possibility of transmission of HIV".

The relevant legislation, s 265(3) Criminal Code RSC 1985, C-46, uses the word "fraud": a complainant does not consent if he or she does not resist or submits by reason of fraud. Here, the defendant did not disclose that he had HIV when he obtained agreement to sexual intercourse. His viral load was low, and in he used a condom with one of the four complainants.

The element of fraud had been interpreted as requiring that the defendant dishonestly deprived the complainant of knowledge about a significant risk of serious bodily harm. Now this risk is explained as a realistic possibility of transmission of HIV.

For the New Zealand law (inaccurately referred to at [50]-[54] of Mabior) see s 128A(7) Crimes Act 1961, which provides that consent is not given by a person who is "mistaken" about the nature and quality of an act.

I expect that this NZ provision, introduced in 2005, will be interpreted to mean that a mistake, in a context like that in Mabior, is a failure to recognise a realistic possibility of transmission of HIV.

McLachlan CJ for the Court stressed that what facts give rise to a realistic possibility of HIV transmission will vary as research increases knowledge. The risk assessments may change. But the evidence in the case disclosed the current understanding of what the relevant risks are.

Broadly, and converting the figures given in the judgment [94]-[101], the evidence was that the probability of transmission of HIV infection during unprotected sexual intercourse may be as high as 0.008 or as low as 0.0005. That is, 8 in 1000, or 5 in 10,000. If a condom is used, the risk reduction is 80%, so at the highest the probability of transmission would be 0.0016. If therapy had reduced the viral load, as it had in this case, to a low level, the risk of transmission may be reduced by roughly 92%, so that if a condom is used the probability of transmission would then be 0.000128. If a condom is not used but the viral load is low, the probability of transmission is 0.00064.

I have just used rough figures as ranges were given for the risk of transmission in unprotected intercourse and for the effect of therapy. But the result of the appeals was that reduced viral load plus use of a condom resulted in a risk of transmission of HIV that was less than "a realistic possibility", whereas reduced viral load but without use of a condom resulted in a risk of HIV transmission that was "a realistic possibility".

That is, "a realistic possibility" is greater than a probability of 0.0001, but a probability of 0.0006 is "a realistic possibility" of HIV transmission.

The context is important: where the consequences are serious the acceptable risk may be lower. If we suppose that the consequences in the HIV transmission context are serious (this may change as therapies develop) we can conclude that the Court is regarding an appropriately low risk of transmission for the purposes of attributing criminal responsibility as one that carries a probability of 0.0006.

How does that compare to life's other risks of death? Well, a few second's Googling indicates that in the UK the average risk of death in any one year from injury or poisoning is 1 in 3137, or a probability of 0.0003. The risk of dying in a road accident in the UK is 1 in 16,800, or a probability of 0.00006. And for cancer death it is 1 in 387, or probability 0.003.

So in Mabior a risk that might have been about ten times the risk of dying in a road accident amounted to a realistic possibility of transmission of HIV. Or about one fifth the risk of dying of cancer.

And, while we are out on this limb, (or "out where the buses don't run", as Kinky Friedman would say), we could use this sort of data as an indication of the level of risk of serious harm that the law considers appropriate for attributing criminal responsibility. Then we should wonder about the probability of wrongful conviction that is sufficient to support a verdict of not guilty. Does it vary according to the harm threatened or caused by the offending? Or is all wrongful conviction equally repugnant?

Don't look at me. How should I know?

For discussion of the meaning of "beyond reasonable doubt" see R v Wanhalla (noted here 25 August 2006, R v JHS, noted here 1 June 2008, and R v Layton, noted here 24 July 2009).