Risk communication is the focus of much misdirected effort for two main reasons, both of which suggest we are better off without the 'r' word'.

One reason is that if risk is not used simply as a synonym for probability it is compounding two entirely different concepts - the probability of a harm (e.g. a fracture) and the undesirability of the harm (the value placed on the consequences of the fracture). The resulting communication, however superficially satisfactory, cannot but be fundamentally flawed as a result of failing to address these separately. On the other hand, if ‘risk’ is used as a synonym for probability, then it would still be better to avoid such use, because of the psychological effects created beyond those already associated with the specified harm that follows the word risk.

Note that we do not hear much complementary talk of 'benefit communication' and this takes us to the second reason for not using ‘risk’:  the necessary decision context of the communication. The communication of a probability (e.g being told that your probability of a fracture in the next ten years is 14%) has no value in itself. It acquires value only through its input into a decision.

Attempting to increase 'understanding' of risk-as-probability, such as via graphic frequencies of faces or figures, is therefore pointless unless it is accompanied by assistance in showing how the resulting ‘understanding’ can be appropriately deployed in a decision. To assume that most of us are able to incorporate a probability such as 14% into a decision - actually a series of such numbers in most multi-option and multi-criterial decisions - is seriously optimistic.

So without interfering or criticising efforts to increase the understanding of probability in general or a specific probability value, and indeed always offering the latter on an opt-in basis, we are dealing with a challenging concept. Probabilities are unique beliefs about the future. Even if they are based on past frequencies they are totally distinct from such frequencies and can never be 'objective'. In this light it is important to offer a reasonable and defensible opinion that reflects the respondent's importance weights, irrespective of their information state or understanding regarding the ratings of the options (which are the probabilities of the options for each criterion). In this case our duty is simply to make clear the provenance of the ratings and the expected value basis of the score that combines those ratings with the respondent's importance weights in arriving at the opinion. This means that for any individual a better decision may not necessarily be an informed decision, if the latter is focused on ‘understanding’ of the probabilities involved.

In the clinical or public health contexts all communication should be decision communication, addressing the probabilities of benefits and harms and the values placed on them, separately, before integrating them into a decision.

For further development of these arguments, see Against Risk, Communication for better decisions: not about ‘risk', and Managing without risk.  See also The Bayesian approach to decision-making.