The Royal Free Journal Club is on a pause for a couple of weeks, so in this hiatus, I’m going to pick up on something that was highlighted in our discussion of Kelly et al’s paper a couple of weeks ago.
In this paper, summarised by Dr Sathananthan , the authors conclude that their result of p=0.06 describes “a trend towards an increased incidence of hypotension following IV paracetamol administration in ICU patients.” However, this is not what the p-value represents at all. A p-value is a representation of the probability that the null hypothesis is rejected, not “a trend towards” anything.
Jeremy Hunt’s misuse of statistics to describe the “weekend effect” has highlighted what can happen if we do not understand statistics or how to interpret them. Andrew Klein over at the Anaesthesia Blog talks about this in more depth. It is our responsibility to ensure that we know how to interpret the data that is being presented to us and not just rely on what the authors or our colleagues say. After all, 50% of anaesthetists are worse than average at interpreting statistics.
by Taj Fregene