How many replicates are necessary to estimate the average defibrillation threshold?-animal experiments and a mathematical model

The ability to defibrillate the heart has been routinely predicted from a single measurement of defibrillation threshold. With any biological variable there is inherent variability, the effect of which can be reduced by obtaining replicate measurements and averaging. Since excessive fibrillation-defibrillation episodes are not clinically acceptable, the authors investigated the minimum replicate number to approximate the mean of seven replicates in 16 anesthetized pigs and 1000 replicates in 100 tests of a mathematical model. Two to three replicates were sufficient to provide accuracy in all animals, while the first measurement was sufficient in the mathematical model.<>