An advisory committee to the FDA said a blood thinner from Bayer and J&J should not be approved for use in patients with acute coronary syndrome.

Sloppy data keeping fueled the majority of doubt among the six panelists who voted against the expanded use of Xarelto (rivaroxaban) Wednesday. That’s despite an upfront endorsement by the FDA panel’s clinical lead reviewer, Caroline Hicks. Four voted in favor of approving the drug for the new indication. Although the lone abstainer, Thomas Fleming, said he withheld his vote because the panel’s goal was to advise, not decide, he noted that missing data caused him to have “important concerns.”

The panel’s 4-6-1 vote does not lock the FDA into a position, but its decision can be taken into consideration.

Among the doubters was Cleveland Clinic Foundation’s Dr. Steven Nissen, who expressed reservations about the quality of data provided by the sponsor, among other concerns. After the vote, Dr. Nissen told MM&M that “the high rate of withdrawal of consent…undermined the ability to interpret the results of the study.”

Citing the number of counted deaths as an example of concerns with the methodology, FDA team member Thomas Marciniak said the data could not be trusted. He took issue with the way Bayer and J&J’s Janssen unit reported different numbers of deaths, first as one, then later saying that the total number of deceased included two more. “This is a major cardiovascular outcome trial,” Marciniak said. “The one stat I hope you have down cold is how many patients died and how many are alive and how many are you uncertain about of their status.”

Marciniak also took issue with the lack of clear, consistent follow-up with patients who dropped out of the study—Bayer/Janssen said 1,294 patients, who were spread across several locations—left the study and the drugmaker was able to confirm 177 were alive. Drugmakers said local privacy laws prevented them from being able to contact everyone but that they did all they could to reach as many patients as possible.

Panelists said drop-outs are as important as participants because they need to know why patients left the study and how they fared after being taken off the medication. Marciniak also found fault with the way in which “contact” was described. The presentation indicated that study sites tried to contact drop-outs but saw active participants in person.

Bayer/Janssen said this was a typographical error and that there was a uniform system for reaching current and former participants. Critics weren’t convinced and said they were concerned that if clinics did not try to see drop-outs in person, it could make them less likely to check-in. They said this is a concern because had these drop-outs been ill, or even healthy, that data would have been excluded from the study. Further, the records did not indicate how patients checked in, which also raises accuracy questions.

“You cannot tell in this study how the patient was seen,” Marciniack told the panel, adding that there was no “indication to indicate who came into the office and who was called.”

Three sites were also excluded from the study because they did not comply with the protocols. This elimination also bothered critics because it’s not just the absence of poor data, but could be an exclusion that could also make the drug’s impact look better.

Xarelto is currently used to prevent stroke in patients with non-valvular atrial fibrillation, and in patients following certain kinds of surgery. The additional use could help it stake out territory in the market formerly dominated by Plavix (clopidogrel), which is currently indicated along with aspirin for the ACS indication. Plavix lost patent protection earlier this month. Company findings indicated that the 2.5-mg dose was preferable to the 5-mg dose. The study said it did increase the risk of bleeding. Considering the bleeding risk, panelist members recommended a black box warning that would keep doctors and patients aware of the potential downside. This possibility did not placate the critics.

“These bleeds are not trivial,” Sidney Wolfe, director of the advocacy group Health Research Group of Public Citizen told the panel when he explained why he voted against the drug’s approval.

Nissen, who is chairman of the department of cardiovascular medicine at the Cleveland Clinic Foundation, told MM&M the other reasons for his negative vote included “marginal statistical evidence of benefit—a ‘p’ value of .039 for benefit at the target dose of 2.5 mg BID—increased risk of intracranial hemorrhage and life-threatening bleeding.” (MM&M will post a Q&A with Dr. Nissen, including more of his thoughts on the Xarelto vote, tomorrow.)

Among the four “yes” votes was acting committee chair Philip Sager, who said the study met the nominal statistical significance it set out to do. He said he voted “yes” based on the risk-benefit profile, yet he had reservations. “The missing data is a real concern both in this study and in ongoing and future trials,” he said.