A reply

Abstract

In their letter, Fabregat-Lopez et al. set out a well-reasoned and spirited attack on the notion that tracheal intubation is the gold standard for airway management. I do not think anything they write is at variance with my own comment in the editorial that “RSI has the weight of ‘historical authority’ (and also has its critics)” [1]. The last line of their penultimate paragraph states a simple fact but I think misses an important point about the role of statistics in hypothesis testing. A statistical test helps reject (or not) a null hypothesis [2]. Here, the null hypothesis (put very simply) is that the device (the Proseal LMA) is ‘ineffective’ (i.e., ‘fails to prevent aspiration’). It is the upper limit of the 95% confidence interval (CI) that tests this notion. The lower limit of the confidence interval is technically irrelevant for this hypothesis test. An upper limit 95% CI of ∼3% (or even ∼1%) for incidence of aspiration does not reject the null hypothesis. So it might reasonably be claimed that Fabregat-Lopez et al. have shown (perhaps rather well) that the device is ‘ineffective’ in this regard. For this reason a zero incidence of anything in 102 patients is really not meaningful for hypothesis testing. I used my editorial to try and show what result or sample size would have been more meaningful [1].