Author + information
- Received September 19, 2011
- Revision received November 11, 2011
- Accepted December 3, 2011
- Published online April 3, 2012.
- Jonathan S. Steinberg, MD⁎ ( and )
- Suneet Mittal, MD
- ↵⁎Reprint requests and correspondence:
Dr. Jonathan S. Steinberg, Valley Health System, Arrhythmia Institute, 5 Columbus Circle, Suite 1500, New York, New York 10019
The federal government has investigated a large number of institutions regarding concerns that implantable cardioverter-defibrillator procedures were performed in violation of the criteria set forth in a National Coverage Determination. We describe our experience and responses to such an audit, as well as the to complexities and nuances of practicing evidence-based medicine in the setting of heavy regulatory oversight.
- CMS indications
- Department of Justice
- federal audit
- implantable cardioverter-defibrillator
- National Coverage Determination
Practice guidelines endorse the use of an implantable cardioverter-defibrillator (ICD) in patients with chronic left ventricular dysfunction (1). Although a significant number of eligible patients are never referred for this life-saving therapy, a recently published study has raised concern that many patients undergoing ICD implantation do not meet these practice guidelines (2).
Over the past year, the Department of Justice (DOJ) has been conducting a civil investigation to determine whether some institutions had submitted claims for payment for implantation of ICDs that were not medically indicated or violated the Centers for Medicare and Medicaid (CMS) payment policy. The period covered by this investigation dates back to 2003 when CMS expanded coverage for primary prevention ICD indications (Table 1). A major focus of this inquiry has been the exclusion of coverage during the 40 days after acute myocardial infarction (MI) or the 3 months after percutaneous coronary intervention (PCI) or surgical revascularization. Although ICDs were the subject of this initial inquiry, dual-chamber pacemaker indications are also being actively scrutinized, and PCI is likely to receive similar attention.
In the recently published study by Al-Khatib et al. (2), using data from the National Cardiovascular Data Registry of 111, 707 primary prevention ICD patients, 22.5% were found to have received an ICD for non evidence-based indications. These included 62% with newly diagnosed heart failure, 37% who were within 40 days of an MI, 12% with New York Heart Association functional class IV heart failure, and 3% within 3 months of bypass surgery (unfortunately, PCI data were not included). The findings reported by the researchers in combination with the DOJ investigations has resulted in a dramatic decline in the number of patients undergoing ICD implantation in the United States.
As electrophysiologists who have been based at an academic teaching hospital, our practice has also directed electrophysiology (EP) programs at neighboring nonteaching facilities. The largest and busiest of our affiliates is a tertiary-care hospital located in a suburban community. This past year, the local hospital administration was informed that an initial analysis of claims submitted to CMS might have included patients who should have been excluded from coverage of ICD implantation. The total number of identified patients was 229, representing 8.7% of all de novo (nonresynchronization) ICD implants for primary prevention during the period of 2003 to 2010 reimbursed by Medicare. Patients were excluded largely because of timing relative to clinical events or interventions but not because of other exclusions, reflecting that this distribution is a function of identification of cases based on coding inconsistencies (Table 2).
A site visit with lawyers representing the DOJ and CMS was scheduled. As director of the EP program, one of us (J.S.S.) was asked to provide expert assistance to the hospital's administrative and legal teams. This narrative will describe the process by which an intense internal review was conducted and adjudicated; how a system of responses to the audit was developed that either conceded excluded implants or attempted to justify other implants despite possible coding transgressions; and how a set of controls was subsequently enacted to prevent any future departures from standards of care and coding and claim submission regulations. Our experience might provide valuable lessons to other institutions, and their physicians, administrators, and lawyers, and, indeed, many colleagues have contacted us to share our thoughts.
An experienced cardiology quality assurance registered nurse performed the initial review of cases. All source material in medical records was examined, and summaries of each patient were entered into a spreadsheet. Particular attention was focused on the alleged coding discrepancy, as patients had been identified based on CPT coding entries that were incompatible with ICD indication. This preliminary review and summary were presented for EP oversight, and it was apparent that many incompatible diagnosis and procedure codes were prevalent. We decided that a complete medical record review would be required to determine whether clinical circumstances dictated technical violation of CMS guidelines, whether coding entries were incorrect, or whether there were genuine non-guideline implantations. Each patient's chart was retrieved, and all relevant components extracted and bundled for detailed review by EP.
During this phase of detailed review, it became evident that a small number of implants were not indicated (n = 34, 1.3% of all implants and 15% of the targeted subgroup); most commonly, these were implanted after bypass surgery in the setting of nonsustained ventricular tachycardia (VT) and/or a positive EP study. On the other end of the spectrum were cases that were clearly secondary prevention but the index cardiac arrest or VT event were documented in the records of another hospital before transfer for the ICD. The vast majority of ICD case indications was more nuanced and highlighted the complexity of adjudicating between clinical practice and the contemporary regulatory environment. These cases could be categorized into 5 groups (Table 3): 1) secondary prevention indication when the presentation was syncope in the setting of cardiomyopathy; 2) concurrent trivial cardiac enzyme leak or enzyme elevation for non-MI reasons but coded as acute MI; 3) ICD implantation when the precipitating acute device need was bradycardia and pacemaker indication; 4) incomplete or incidental percutaneous revascularization not anticipated to have any meaningful effect on chronic LV dysfunction; and 5) ICD implantation near the end of the 90-day post-revascularization period when the patient was admitted for heart failure. These categories represented real-life situations encountered in our care of patients, and we believed that ICD implantation was generally justifiable, even at times in violation of the NCD directives.
Unexplained syncope in a patient with severe LV dysfunction is usually due to sustained hypotensive VT and is considered a very high risk clinical event that warrants secondary protection by an ICD even if VT is not clinically documented (1).
The advent of high-sensitivity cardiac enzyme assays has made management of acute coronary syndrome more informed, but these enzymes are also positive in a variety of clinical settings not indicative of acute MI, including heart failure exacerbation, atrial fibrillation, cardioversion, and noncardiac events. In addition, a patient with long-standing ischemic cardiomyopathy with well-documented severe LV dysfunction who presents with a small enzyme leak from ischemia is very different from the patient with an initial presentation of acute ST-segment elevation MI and positive enzymes. Indeed, the NCD also indicates that positive enzymes alone are not a criterion for the diagnosis of an acute MI. These former subsets of patients may have been recognized as “MI” for coding purposes but from a clinical perspective are managed quite differently and have a prognosis including risk of sudden cardiac death governed by the pre-existing LV dysfunction.
There is a need for pacemaker implantation in a small proportion of acute MI patients and cardiac surgical patients for bradycardia indications. If these patients have had severely reduced LV function in the past, physicians are faced with a serious quandary: implant a pacemaker and upgrade it to an ICD when the exclusion period of 40 days or 90 days expires, or implant an ICD upfront to serve the dual purpose of treating bradycardia and long-term risk of serious ventricular tachyarrhythmias. The first approach, which does not violate the NCD exclusion criteria, exposes a large percentage of patients to the risks of an upgrade procedure that are actually greater than primary implant surgery. The latter approach, which may violate the NCD exclusion criteria, will expose a small percentage of patients to an unnecessary ICD if LV function improves sufficiently during the waiting phase. This decision analysis is highly representative of the possible conflicts that exist in day-to-day practice, given reimbursement regulations; we have often opted to implant the ICD rather than the pacemaker on the grounds that it is in the patient's best interests when the clinical context virtually guarantees that LV function will remain depressed.
Another frequently encountered clinical scenario is the patient with known cardiomyopathy who is admitted for heart failure exacerbation. Many of these patients will undergo a coronary angiogram, and if an obstructive lesion is identified, will almost always have a PCI performed. Because these patients have well-established ventricular dysfunction and because the coronary intervention is not in the setting of ST-segment elevation MI, it is highly unlikely to substantially improve LV function (3). Moreover, the necessity for heart failure hospitalization places the patient in a very high risk category for near-term cardiac events, including arrhythmic death (4). We have often attempted to provide maximum protection for these patients with an ICD. Alternative approaches such as deferred ICD for another 3 months or use of a wearable defibrillator avoids the appropriate clinical intervention simply to satisfy administrative constraints.
Finally, the 90-day exclusion period after revascularization cannot always be considered an inviolable barrier to ICD implantation. It is not in the patient's interest to defer protection of an ICD when the expiration of the 90 days is imminent and the patient is in a very high risk category, such as heart failure decompensation. It is intuitive that nothing will change in the patient's risk profile if a few more days elapse, and of course sudden death can strike at any moment. Furthermore, although we know that ICDs are not effective within 40 days of an acute MI (1), we really do not know the role of the ICD in the first 3 months of heart failure diagnosis or post-revascularization because these patients were excluded from clinical trials. Thus, there is sometimes a lingering sense of concern that some of these patients will die suddenly while waiting for an ICD.
We found the government legal team highly knowledgeable and informed. The lawyers listened thoughtfully to our presentation about the results of our review and how we viewed the balance between complying with the letter of claims regulations and the subtleties of clinical practice. They were quite sensitive to avoiding situations that could present harm to the patients simply to satisfy the coding guidelines and, in large part, were receptive to the “exceptions” that are detailed in the preceding text.
In response to the arduous review process after the DOJ inquiry, we made several important changes. Coders have been educated that not every patient with positive cardiac enzymes is coded as an MI, and other clinical criteria and review mechanisms were implemented. Each ICD implant undergoes a concurrent peer review during a morning conference call of practicing EPs. The implanting EP must also complete a form at the implant procedure that requires identification of the indication for ICD and that all exclusions are absent. Mitigating circumstances, as defined previously, must be included and detailed if present. Finally, a post-hoc review is routinely undertaken by quality assurance nursing staff, and all questionable implants are referred for additional independent physician review.
Compliance with reimbursement regulations is almost certainly variable, although our experience suggests a more complex dynamic based on clinical context. It is also suggested that reimbursement regulations are often in conflict with more flexible, nimble, and updated published guidelines. The latter are based on evolving medical evidence and created by committees of experts. It is these published and vetted pathways that have precedence in guiding physicians regarding who requires an ICD and when it should be implanted. The NCDs should be updated to reflect these dynamic documents or consider ceding authority to them.
Dr. Steinberg is a consultant for Medtronic, St. Jude Medical, and Cameron Health, and has received research support from Medtronic. Dr. Mittal is a consultant for Medtronic, Boston-Scientific, and Biotronik.
- Abbreviations and Acronyms
- Centers for Medicare and Medicaid Services
- Department of Justice
- implantable cardioverter-defibrillator
- myocardial infarction
- National Coverage Determination
- percutaneous coronary intervention
- ventricular tachycardia
- Received September 19, 2011.
- Revision received November 11, 2011.
- Accepted December 3, 2011.
- American College of Cardiology Foundation
- Epstein A.E.,
- DiMarco J.P.,
- Ellenbogen K.A.,
- et al.