More Studies of Investigational High-Risk Cardiovascular Devices are Getting Published

Nearly 90% of the pivotal studies conducted to secure FDA approval of 35 novel, high-risk cardiovascular devices between 2011 and 2013 were published in the peer-reviewed literature, a new analysis shows. 

Take Home: More Studies of Investigational High-Risk Cardiovascular Devices are Getting Published

That’s up from a publication rate that hovered at around 50% in the first decade of the 21st century, according to Adam Phillips, MD (Yale University School of Medicine, New Haven, CT) and colleagues, writing in a research letter published online ahead of print in JAMA Internal Medicine.

“In 2007, the FDA Amendment Act was passed expanding the registration and reporting requirements for studies on and it explicitly included medical devices,” Phillips told TCTMD. “We know from some other studies in the drug world that closer to nine out every 10 trials are published, so we thought there might be some improvement in the publication rate of high-risk cardiovascular devices, which account for a very high proportion of the high-risk devices that are approved by the FDA.”

Indeed, Phillips added, roughly 50% of high-risk devices approved by the FDA are for cardiovascular indications.

Phillips and colleagues searched the FDA’s premarket approval database for all novel, high-risk cardiovascular devices approved between January 2011 and December 2013, identifying 35 devices. Next they reviewed FDA documents and additional publications to find any clinical trials supporting device approval and searched

They found 70 studies conducted to support approval, of which 56 were published in the peer-reviewed literature; an additional five studies were reported on Zeroing in on pivotal trials, 34 out of 38 studies (89%) were published, whereas only 22 out of 32 nonpivotal studies (69%) were published.

Of note, in a comparison of results in the medical literature versus those on, Phillips et al found three studies for which the primary outcome results were discordant: the STROLL study of a nitinol self-expandable stent for obstructive superficial femoral artery disease; the Zilver PTX drug-eluting stent single-arm study; and the EVEREST II randomized controlled trial of the MitraClip. Of these, only STROLL, which used a more lenient definition of the primary patency endpoint in the published literature than in those reported to the FDA, would have altered the public interpretation of the results, the authors say.

Asked about this discord, Phillips said there was no particular signal of pattern or intent among the three studies. “It’s really hard to glean anything from such small numbers,” he said.

Among other insights from the analysis:

  • Studies of devices designated as life-sustaining were more likely to be published than those for devices designated non-life-sustaining (94% vs 66%; P = .004).  
  • Publically funded studies were more likely to be published than those funded by private companies (87% vs 56%; P = .006). 
  • There was no significant difference in the rate of publication among studies granted priority review vs those following the standard FDA review, but numbers were small (10 out of 13 studies and 46 out of 57 studies, respectively).

The key point, Phillips stressed, is that publication rates for these high-risk device studies have increased, which suggests that the FDA Amendment Act had the desired effect. The “goal” he added, would be 100% publication, but that “may be a ways off,” he said.

As reported by TCTMD, another recent analysis, also out of Yale, showed that nearly one-third of clinical trials conducted at 51 academic medical centers in the United States remain unpublished within two years of completion.

Phillips AT, Rathi VK, Ross JS. Publication of clinical studies supporting FDA premarket approval for high-risk cardiovascular devices between 2011 and 2013: a cross-sectional study. JAMA Intern Med. 2016;Epub ahead of print. 

Related Stories:


  • Phillips reports no relevant conflicts of interest.

We Recommend