One of the challenges in seeking evidence to aid a difficult clinical dilemma is decided what the questions which are amenable to study might actually be. There are elements of the triad of clinical expertise, best-available research and patient situation and preferences which can only be brought by the individuals.

There was a quote from a Cochrane Conference recently about the benefits of patient/public involvement in research: “’Cause it might reduce the chance of the researcher making an absolute arse of themselves”. Those with a condition may have a much clearer idea than a researcher, or clinical doctor, about what needs to get better. It’s not just researchers who need to heed this advice. If you’re going to practice in an evidence based way, you’re going to have to ask the patients and families where their priorities lie. Take nausea and vomiting, for example. Mostly, you’d think that complete removal of these symptoms was a sensible aim. However in patients who have intractable, terribly difficult to control vomiting, where the only therapies working are those which keep them asleep 22/24 hours, this may not be the case.

There are other elements where the science feels like it should be able to help. For example the question may arise “Which toddlers with wheeze should we prescribe oral steroids for?”. This isn’t quite the question “Does prednisolone improve symptoms in pre-schoolers with viral induced wheeze?” and it’s not quite “Which pre-schoolers will go on to develop atopic asthma?”, but it certainly nods to both.

How this type of question can be addressed could be by looking at subgroups of large trials, to see if there are those who are predictable ‘responders’. Beware, however, the perils of p-fishing and the astounding ability of us medical types to come up with a good reason why something is plausible for almost any observed effect. Be extra-wary when a systematic review draws a sub-group meta-analysis: examine closely for selective outcome reporting bias. It may be you end up addressing the issue with recourse to indirect data and assumptions; pred good for asthma, risk factor F is good to show who will get asthma, so give pred to those with risk factor F.

Each approach to a challenging question will have some strengths and some difficulties. Your job in undertaking an appraisal is to explore those and still make the best judgement you can.

• Archi