Using registries to assess clinical practice and improve patient care

An article published in the January 5, 2011, issue of the Journal of the American Medical Association (JAMA), "Non-Evidence-Based ICD Implantations in the United States," stirred widespread comment in the medical and general press.

The research assessed patients receiving an ICD for primary prevention of sudden cardiac death. The patients were entered in the National Cardiovascular Data Registry (NCDR) ICD Registry between January 1, 2006, and June 30, 2009, to determine if they received the ICD on the basis of established practice guidelines. The study reported that 22.5% of patients did not meet evidence-based criteria for implantation.

It is important to understand how the ICD Registry was developed and how this article highlights the emerging practice of measuring and public reporting of physician and hospital outcomes, with a goal of improving the quality of medical care.

Developing the ICD Registry

The registry was developed through a partnership of the Heart Rhythm Society (HRS) and the American College of Cardiology Foundation (ACCF) using the expertise of the NCDR. The registry was mandated by the Centers for Medicare & Medicaid Services (CMS) in the National Medicare Coverage Decision to expand ICD coverage on the basis of the Sudden Cardiac Death in Heart Failure Trial (SCD-HeFT) results. The National Medicare Coverage Decision described a policy termed Coverage With Evidence Development that allowed expanded coverage as long as patients were entered into a registry to track outcomes.

The registry began collecting data in April 2006 and now includes all 1,489 hospitals in the United States performing ICD procedures. It has collected data on more than 750,000 ICD placements. "The CMS directive is to enter data on Medicare beneficiaries receiving an ICD for the primary prevention of sudden cardiac death, but to their credit, 84% of hospitals have chosen to submit data on all device recipients, regardless of age or device indication," according to Stephen C. Hammill, MD, a cardiologist at Mayo Clinic in Rochester, Minnesota, and an author of the JAMA paper.

Approximately 95% of all ICD procedures in the United States are entered into the ICD Registry. This extensive reporting provides the most comprehensive characterization of contemporary ICD practice and permits meaningful comparison with published randomized controlled trials such as SCD-HeFT and the Multicenter Automatic Defibrillator Implantation Trial II (MADIT-II).

Characteristics of patients entered in randomized controlled trials and the ICD registry

Enlarge

The Table lists the characteristics of patients enrolled in SCD-HeFT and MADIT-II compared with patients in the ICD Registry, indicating that the registry patients are older, with a greater proportion of women and patients with atrial fibrillation, hypertension, and diabetes. Thus, the ICD Registry better depicts the types of patients receiving ICDs in the real world, in contrast to randomized controlled trials where patient entry was restricted. The registry has made substantial progress toward several predefined goals:

• Reveal the degree to which clinicians are managing ICD therapy in accordance with evidence-based medicine.
• Enable clinicians to compare their in-hospital outcomes with those of other physicians.
• Provide insights for clinical investigation.
• Highlight the ICDs' performance outside clinical trial constraints.
• Provide a detailed view of the morbidity, mortality, and resource utilization associated with ICD therapy.
• Assess local hospital needs for quality assurance and quality improvement.
• Serve as a hospital and physician response to performance measures initiatives.

"A key aspect of the registry is to improve quality performance at hospitals implanting ICDs, which is achieved, in part, through benchmarking reports provided to hospitals on a quarterly basis, detailing the outcome for all data elements and summarizing performance metrics," says Dr Hammill.

Each hospital is compared with hospitals of similar procedure volume and the national aggregate. Reviews of the annual data reports published by the ICD Registry Steering Committee since 2007 have demonstrated a gradual trend in improvement of outcomes during the first 4 years of registry activity. The total procedure-related adverse events have decreased from 3.77% in 2006 to 2.87% in 2009.

The ICD Registry is part of the NCDR program, which assesses outcomes in multiple different areas of cardiovascular disease, including cardiac catheterization and percutaneous intervention, carotid artery stenting, and outpatient cardiovascular practice. Data from the ICD Registry have been used in several important publications, providing information on patients receiving ICDs that was not available through the randomized controlled trials, including:

• Understanding the application of ICD therapy in the general population
• Gender differences
• Procedure-related adverse events
• Association of physician certification to outcomes and appropriate use of cardiac resynchronization therapy with defibrillator (CRT-D)
• Racial and ethnic differences in CRT-D use
• ICD outcomes in patients with end-stage renal disease
• The relationship between hospital procedure volume and complications of ICD procedures
• The prevalence and predictors of off-label use of CRT-D

Brief review of the JAMA article

The JAMA publication reported that 25,145 of 111,707 ICD placements (22.5%) were non-evidence-based and were associated with a significantly higher risk of in-hospital death and procedure-related complications. Patients were classified as receiving a non–evidence-based implant if they had:

• Myocardial infarction within 40 days prior to ICD placement
• Coronary revascularization within 3 months prior to ICD placement
• New York Heart Association (NYHA) class IV symptoms
• Newly diagnosed heart failure at the time of ICD placement

National guidelines developed by the ACCF, American Heart Association, and HRS have described implantation criteria that were developed on the basis of results of randomized controlled trials. In addition, CMS will not reimburse for ICDs placed in the patient groups listed. Of all the patients who received ICDs:

• 8.3% were implanted within 40 days of myocardial infarction
• 0.73% were implanted within 3 months of coronary revascularization
• 2.7% were implanted in patients with NYHA class IV symptoms
• 14.0% were implanted in patients with newly diagnosed heart failure

The ICD Registry data form asks if the patient has a history of congestive heart failure and, if so, was it within the past 3 months, 3-9 months, or greater than 9 months. This qualifying question is in the background of the guidelines recommending patients be on maximal medical therapy before receiving the ICD, and it is assumed to be unlikely that maximal medical therapy can be achieved in patients with symptomatic congestive heart failure with less than 3 months of treatment.

"This study and others documented that the rate of non-evidence-based implants was significantly lower for electrophysiologists, who have been shown in prior publications to have fewer complications and more appropriate device selection than other physicians performing the procedure," says Dr Hammill.

It is important to note that many patients appropriately receive ICDs outside guidelines on the basis of the physician's clinical judgment and the patient's presentation. This group includes patients who have an established cardiomyopathy and present to the hospital with a small troponin leak subsequently coded as a myocardial infarction, although the acute event was not the cause of the cardiomyopathy, which preexisted the small myocardial infarction.

In addition, patients may require coronary revascularization and have either a preexisting cardiomyopathy or an indication for a permanent pacemaker. In such patients, it may be elected to place the ICD because of their increased risk of sudden death and need for a pacing device.

Public reporting of outcomes

CMS and other payers are increasingly interested in reporting valid measures of patient outcomes. This reporting requirement has been put into law by the US Congress. Unfortunately, administrative databases developed by payers to assess provider outcomes have several limitations:

• Data definitions are often imprecise.
• Final coding may not be supported by the clinical record in a substantial proportion of cases.
• It is difficult to distinguish comorbid conditions from complications.
• Important clinical risk data such as ejection fraction and NYHA symptom classification are not available.

Prospective clinical registries maintained by professional societies, such as the ICD Registry, eliminate the inherent deficiencies of administrative data. While the clinical NCDR registries are more detailed and accurate than administrative data, they are limited by the lack of long-term follow-up. Obtaining reliable follow-up information using chart level data or subsequent patient contact is too costly and resource intensive to collect in a representative national sample as large as that included in the ICD Registry.

A hybrid approach being used to develop reliable performance measures combines NCDR clinical data with Medicare Claims Data for follow-up, thus capitalizing on the strengths of both data resources. CMS is working through the National Quality Forum (NQF) and professional societies, including HRS and ACCF, to develop hospital and physician performance measures that will be publicly reported.

One such performance measure to assess outcomes following ICD placement using data from the ICD Registry and Medicare Claims Data has been approved by the NQF. The initial analysis using this performance measure identified the median complication rate for a hospital was 7%, with the lowest decile being 4% and the highest decile 13%. This wide range of complications provides an opportunity for improvement by moving hospitals with the highest rate of complications closer to the median and moving the median closer to the lowest decile group.

Rates of non-evidence-based implantable ICDs

Enlarge

The recent JAMA article similarly showed a wide variation, with some hospitals implanting only 5% of ICDs outside guidelines and other hospitals implanting 50% of ICDs outside guidelines. Ralph Brindis, MD, MPH, president of the ACCF, stated that "when we see that level of variation, there is no way even a skeptic could say that we don't have room for improvement in the way that we apply ICD technology."

Although not addressed in the JAMA article, a major clinical problem is that approximately 50% of patients who meet guideline indications for ICDs do not receive the device. "This situation suggests that physicians are not following practice guidelines and are failing to refer many patients who would benefit from this lifesaving therapy," says Dr Hammill.

The federal government is developing incentive programs for adoption of electronic medical record technology that may provide physicians with the ability to do clinical decision making and share decision responsibilities with their patients, thereby reducing the guesswork and improving the practice of evidence-based medicine. Cardiology has led the medical profession in developing practice guidelines based on well-designed randomized controlled trials. These guidelines are used to aid clinicians in practicing evidence-based medicine. Registries then demonstrate to physicians how the guidelines are being applied and evidence-based medicine is being practiced in the community.

Points to remember

• The implantable cardioverter-defibrillator (ICD) Registry better represents patients in the community, in contrast to clinical trials.
• Of Registry patients, 22.5% did not meet evidence-based criteria for ICD implantation.
• Wide variation in implantation rates outside guidelines occur between institutions
• Many patients appropriately receive ICDs outside guidelines based on clinical judgment and patient presentation.
• Approximately 50% of patients who meet guideline indications for ICDs do not receive the device.