Matched-pairs defibrillation efficacy hypothesis test

It is common to statistically test the hypothesis that one defibrillation technique is better than another. A standard approach is to estimate the DF50 (the shock strength which defibrillates 50% of the time) on a set of subjects under test and control conditions, analyzing the results by pairing the DF50's on an animal-by-animal basis. This paper presents a new approach based on pairing the data on a stimulus-by-stimulus basis. The new approach (matched-pairs) is shown to be easy to use, more versatile, and more sensitive (lower p-values) than the standard approach. The matched-pairs approach is particularly useful in clinical studies where many subjects may receive only a few stimuli.