We’d all like to know what will happen in the future. Well. In some circumstances. For some people. Sometimes. (I’d personally like to know where the Leeds Rhinos will finish next season, but less happy about knowing if and when I’ll become a Grandfather.)

But this basic idea, knowing the answer to “If I do Thing B will L occur?”, is something we’d all like to be able to tell our patients. It seems even more prominent a need when the Thing we are about to do is a one-off, major occurrence, like an operation or a bone marrow transplant. If we are ‘just’ giving a medicine, or an ointment, we can stop of change or alter what we’re doing more easily. The challenge, in terms of getting the evidence to underpin such decisions, is the difficulty in sometimes undertaking randomised trials in these areas.

So what can we do? We can use well constructed prediction models, where we look for ‘prognostic factors‘, and maybe fancy statistical approaches with ‘propensity scores‘ or ‘causal analysis‘. We can make sure we do draw what we can from randomised studies, extrapolating and extending the info we have. And very importantly, we can avoid being drawn into the mythologies of purely pathophysiological discussions supported by a single study with a significant p-value. Using this evidence and our appraisals of it will allow us then to ‘do’ evidence based practice better. We can make sure when we talk with families to be exceptionally clear that all our predictions are somewhat uncertain. As with all of our conversations, we can consider adding the “What if ..?” question too, exploring what if the therapy doesn’t work, and where else we might go.

Making big decisions is difficult, but having the best quality evidence to underpin them improves our chances of doing them correctly.