ICD Use Curbed at Hospitals After US Department of Justice Scrutiny

The controversial investigation, first announced to hospitals in 2010, seems to have had its intended effect on use of the devices for primary prevention.

ICD Use Curbed at Hospitals After US Department of Justice Scrutiny

Sparked by whistleblowers’ concerns about misuse of implantable cardioverter-defibrillators (ICDs), a US Department of Justice (DoJ) investigation into the issue, first announced to hospitals in 2010, seems to have had its intended effect.

Hospitals that settled with the agency due to the investigation saw sharper decreases in ICDs for primary prevention compared with those that didn’t settle, researchers have found. Moreover, secondary prevention with ICDs held relatively steady, and patterns didn’t differ based on whether patients were or weren’t Medicare beneficiaries.

Back in 2005, the US Centers for Medicare & Medicaid Services (CMS) issued a national coverage determination (NCD) for ICDs specifying, among other things, that implantation of the devices should be delayed until 40 days after MI and 90 days after PCI or CABG. In 2008, two individuals filed a formal complaint under the False Claims Act that these waiting periods often went unheeded.

The DoJ launched its investigation that year, only announcing it to hospitals in 2010 and to the public in 2011. Ultimately, “more than 500 hospitals settled with the DoJ for more than $280 million,” with the whistleblowers taking home a cut of $41.8 million for their efforts, according to an editorial by Paul A. Heidenreich, MD (Stanford University School of Medicine, Palo Alto, CA), published alongside the new paper this week in JAMA.

Senior author of the study, Jeptha P. Curtis, MD (Yale School of Medicine, New Haven, CT), told TCTMD that the patterns seen in ICD use show “that the investigation seemed to do quite effectively what it was intended to do, which is to make sure that clinicians implanting these devices paid very close attention to the criteria for reimbursement. And speaking as a clinician, sometimes we don’t think about that when we’re taking care of an individual patient, not that it should be the only factor, but I think it needs to be an important one.

“Seeing the rapid change in the proportion of cases that weren’t meeting the national coverage decision is in fact encouraging—that people responded in a positive fashion,” he said. “The second encouraging finding is that we really didn’t see any drop in the volume of secondary-prevention ICD placements. And that’s something going into the study we were really worried that we would find.”

Physicians and hospitals indeed change what they do under the purview of regulation and law. Mintu Turakhia

Mintu Turakhia, MD (Stanford University and VA Palo Alto Health Care System), commenting on the research for TCTMD, said it is revealing.

“It was known that the DoJ investigation identified bad actors and recovered costs. What was not known is whether the DoJ investigation led to behavior change,” he noted via email. “This study proves that it did—not just at the audited hospitals, but across all hospitals participating in the ICD registry. Physicians and hospitals indeed change what they do under the purview of regulation and law.”

Tracking ICD Use

To chart variations in ICD use over time, Curtis and investigators, including lead author Nihar R. Desai, MD (Yale School of Medicine), looked at 300,151 initial primary prevention ICDs implanted in Medicare beneficiaries between 2007 and 2015 at 1,809 US hospitals in the National Cardiovascular Data Registry ICD registry. One-quarter of those hospitals reached settlements with the DoJ related to ICD use not meeting NCD criteria.

In the first 6 months of 2007, the proportion of ICDs not meeting these standards for reimbursement was higher among hospitals that later reached settlements than among those that did not (25.8% vs 22.8%; P < 0.001). Over the 9-year study period, the use of ICDs that didn’t align with the NCD decreased in both groups, but the decline was more pronounced in the settlement hospitals (P for interaction < 0.001).

Changes in ICDs Not Meeting NCD Criteria: 2007-2015




P Value

Relative Decrease



< 0.001

Absolute Decrease



< 0.001

However, ICD use significantly diverged between hospitals only in the period of January 2010-July 2011, which encompassed the time after hospitals were first notified of the DoJ investigation. Settlement hospitals saw “larger and more rapid decreases” than nonsettlement hospitals during these months, Desai and colleagues note.

Specific to primary prevention, ICD use that didn’t meet NCD criteria dropped by a relative 32.8% at settlement hospitals and a relative 27.4% at nonsettlement hospitals over the entire study period (P for interaction < 0.001). Meanwhile, use of ICDs for secondary prevention saw more modest changes, and ICDs among non-Medicare beneficiaries followed similar trajectories.

The researchers point out that the proportion of ICDs not meeting the NCD standards had already started to modestly decrease between January 2007 and December 2009.

The DoJ’s involvement seems to have hastened changes, Heidenreich observes in his editorial. “The mere announcement of the investigation appeared to have a large and immediate influence on prompting hospitals to limit inappropriate ICD implantation for primary prevention. . . . As a form of audit and feedback, the DoJ investigation appeared to be highly effective in changing practice,” he writes.

“Past studies of audit and feedback show relatively modest effects on changing physician behavior, although these studies did not involve allegations of fraud with financial penalties. Clearly, the reward or penalty attached to the feedback influences clinician behavior, with penalties likely more effective in promoting change,” Heidenreich says.

Subsequent to the investigation, interest in appropriate use criteria has grown and the NCD for ICDs has been updated. As of this year, the NCD is now better aligned with clinical guidelines and includes exceptions to waiting periods as well as a requirement for shared decision-making, he notes.

For Curtis, the revised NCD is indeed an improvement over its predecessor. In particular, he said, the shared decision-making aspect “reflects a larger movement towards requiring [these] tools in clinical practice. I think this feels uncomfortable to clinicians to do that, and there’s definitely resistance to using shared decision-making tools, but I think they’re here to stay and we’re going to have to figure out how to work that into our routine clinical practice.”

That the NCD specifies shared decision-making must be documented for ICDs is in some ways concerning, Turakhia said. “Shared decision-making should be used for all decisions, so documenting them for some procedures and not for others gives a message that doctors can make some decisions but not others on behalf of patients—which makes no sense to me.”

‘Letter of the Law’

Investigations into healthcare fraud will continue, Heidenreich predicts. “With big data available to an increasing number of analysts, sizable financial incentives, and growing reliance on appropriate use criterion, more whistleblower complaints may be anticipated,” he concludes. “However, physicians and hospitals can lead (rather than be chased) by developing appropriate use criteria and participating in registries tracking appropriate care.”

I think we can say that it had an effect, and again it’s probably in the eye of the beholder, but I think overall [it was] more positive than not. Jeptha Curtis

Indeed, back in 2010, the DoJ’s move “was not received warmly by the clinical community at large,” Curtis said. “There was a lot of hand-wringing and fear that this represented an unwelcome intrusion into clinical care by the government, and you could understand how people would feel that way. Now with the perspective of 8 years after the announcement of the investigation and it having been wrapped up, I think we can say that it had an effect, and again it’s probably in the eye of the beholder, but I think overall [it was] more positive than not.”

Before, “people weren’t necessarily following the letter of the law,” Curtis explained. “What the government asserted with this investigation is that that the letter of the law matters and that we as clinicians need to be . . . very vigilant that we’re justifying in a very detailed fashion the reasons why we’re performing these procedures. And that is a trend that is not going to go away.”

Turakhia, too, said that medicine has been changing. While there was “some resentment and frustration” over the idea of ICD use being scrutinized, “this was roughly 10 years ago—at a time when physicians were used to having an unchecked level of autonomy, with much of their decision-making buried in paper charts,” he pointed out. “Over the last 10 years, we have seen complete adoption of [electronic health records] public reporting of quality measures—and their linkage to reimbursement.

“So I believe the field has accepted greater accountability and transparency, and that is a good thing,” he concluded.


  • This work is supported by a grant from the Agency for Healthcare Research and Quality (Desai). This research was supported by the National Cardiovascular Data Registry (NCDR). The analytic work for this investigator-initiated study was performed by the Yale Center for Outcomes Research and Evaluation Data Analytic Center with financial support from the American College of Cardiology.
  • Curtis reports receiving salary support from CMS and NCDR and having equity interest in Medtronic.
  • Desai reports being the recipient of a research agreement from Johnson & Johnson through Yale University to develop methods of clinical trial data sharing and working under contract with CMS.
  • Heidenreich reports no relevant conflicts of interest.
  • Turakhia reports serving as a consultant to Medtronic and Abbott.

We Recommend