Developing normalized strength scores for neuromuscular research.

summary scores created from normalized data con-trols for these differences, reduces variance andminimizes the chance of finding false-positiveresults, as fewer variables are analyzed. 27,28 Summary scores are routinely used to analyzeTQNE data. The 20 raw MVIC values are first con-verted to z-scores based on a population mean andstandard deviation for each muscle group derivedfrom a natural history ALS data bank. Two re-gional megascores are then calculated by averagingthe 10 individual leg z-scores (megaleg) and the 10individual arm z-scores (megaarm). This methodplaces all muscles on a ‘‘common yardstick,’’ thusallowing each muscle, regardless of size, to contrib-ute equally to the megascore. Disease progressioncan then be expressed as the slope of the mega-score over time (megaslope). 29 Unfortunately, megascore values are expressedas z-scores and, thus, clinical interpretation is diffi-cult. In addition, megaslope analysis can be mis-leading, because the slopes are often very small frac-tions. Therefore, clinically insignificant changes canresult in a several-fold change in slope (e.g., the dif-ference between a megaslope of 0.001 and a meg-aslope of 0.0001 is not clinically detectable,although, statistically, the difference is tenfold).An improved method for summarizing TQNEdata is to convert raw MVIC values to percent of pre-dicted normal (PPN).