"Any time you do a study, there are certain data that are considered outliers," Gupta acknowledges. "However -- and this is a big 'however' -- in almost all good studies, that is a very small percentage of the total sample size. ... Whenever a significant percentage is excluded, for any reason, it leaves a reviewer skeptical of the data analysis."
Colamarino says those patients spent more time in the operating room, lost slightly more blood and suffered a higher rate of adverse events than other participants in the study. However, she says, they actually had better clinical outcomes than the other patients based on two measures. She also stresses that Johnson & Johnson supplied the FDA with safety data on the training patients during the approval process.
Still, Brent Blumenstein -- a voting member of the FDA advisory panel -- specifically complained about omitted patients during the Charite hearing.
"The sponsor's definition of the ITT (intent-to-treat) population ... is incorrect because it deletes randomized patients," Blumenstein said. "To modify the definition of ITT or analysis by arm by deleting patients is tantamount to saying someone is only partly dead."Like the rest of the panel members, Blumenstein nevertheless went on to vote for approval of the device anyway.