No abstract

Dr Woodcock's point about the effect of the failure rate had been considered. In any experimentation on human subjects missing data from subject loss tend to occur; there is no way of completely overcoming this problem, either clinically or statistically. We believe our analysis provided the most reasoned assessment of the data by (1) analysing between group failure rates and (2) analysing the existing data while explicitly acknowledging the possible effects of missing data. Furthermore, both sets of data are presented and both show small positive effects for antimicrobials, so there is little chance of the reader beirng misled.

No abstract