In Pursuit of Value: Truths, Myths & Urban Legends About Data Registries

The explosion of cardiovascular data registries, and their vast research output, has become the major proof point for field’s commitment to continuous quality improvement (QI). Now, registries have been assigned another purpose. Positioned front and center in the Medicare Access and CHIP Reauthorization Act’s (MACRA) proposed rule, registries that garner the approval of the Centers for Medicare & Medicaid Services (CMS) also may impact reimbursement calculations.

Measurement & management

Over the last three decades, registries have been launched in many cardiovascular subspecialties. The American College of Cardiology  (ACC), with various partners, runs 10 registries focusing on therapies as well as cardiovascular-related conditions. Registries also are the QI centerpieces for other medical societies, such as the American Association of Cardiovascular and Pulmonary Rehabilitation, American Society of Nuclear Cardiology and Society of Thoracic Surgeons.

These organizations promote their registries to health providers and institutions as tools for improving performance on quality metrics; however, registries also have been recognized as data troves that researchers can mine—at much less cost than conducting randomized clinical trials (RCTs)—to uncover sometimes-controversial truths about real-world cardiovascular practice.

“The NCDR robust datasets hold answers to complex questions related to patient risk factors and outcomes; procedure and treatment trends; guidelines adherence; and device, facility and provider characteristics,” promises the ACC-National Cardiovascular Data Registry (NCDR) website, where nearly 900 NCDR-derived published manuscripts and presented abstracts are listed. The answers to such questions are the key to better management of cardiovascular disease, says Ralph G. Brindis, MD, MPH, the NCDR’s senior medical officer for external affairs. “We can’t manage things unless we measure them,” Brindis said when he joined Cardiovascular Business for a conversation about the future of registries and how they are perceived.

What do you wish physicians knew about registries?

First, there’s the incredible bandwidth of registries for patients and physicians. Registries support opportunities for clinicians and hospitals to improve their own quality of care. This is even more true today than in the past, as we transition from the SGR [sustainable growth rate], which rewarded volume, to MACRA, with its focus on value and reimbursing physicians for the quality of our work. The NCDR’s PINNACLE and CathPCI registries are already approved by CMS as Qualified Clinical Data Registries (QCDRs), which means they will meet the needs of physicians who will be reimbursed under MIPS [the Merit-Based Incentive Payment System outlined in the MACRA proposed rule]. Not only will clinicians who are enrolled in these registries avoid penalties but they’ll also be positioned to earn greater reimbursement.

Another reality of our times is that consumers, payers and purchasers expect increased transparency of the clnical outcomes of the care that we provide. Much of the public reporting data available today are from sources that are not transparent to clinicians and other external stakeholders. In fact, it’s possible for a physician or hospital to be ranked high in one system and poorly in another, which underscores the need for public reporting based on true clinical data, as is most accurately found in registries.

Another often-overlooked benefit of registries is that they allow physicians to deliver innovative therapies much earlier than in the past. An example is TAVR [transcathether aortic valve replacement], where data in the ACC’s TVT registry allowed the FDA to expand clinical indications for TAVR use much earlier than would have been possible if the FDA had required an RCT.

What should policymakers and patients know about registry data?

Policymakers understand registries, but their challenge is that cardiology is so far ahead of the rest of medicine. For them to have broad-based policies, particularly surrounding payment reform focused on quality of delivered care that is well captured in our NCDR, won’t be reasonable until other specialties catch up.

Additional value for the FDA and patients is in the potential for registry-based sentinel systems to pick up device- and drug-related safety issues. In many cases, previous methodology for post-market surveillance was not well-positioned for picking up signals of low-frequency adverse events. The best example is the defective Fidelis (Medtronic) ICD [implantable cardioverter-defibrillator] lead that was recalled in 2013. If there had been a registry-based sentinel system in place, such as we now have available with the NCDR, the problem with Fidelis could have been identified about 25 months earlier, and as many as 70,000 people might not have received defective leads. This would have been a service to industry, the FDA, payers, purchasers and, most of all, patients. That’s huge patient value.

The other thing is that registries inform our clinical practice guidelines, which are typically based on RCTs where the average age is typically 62 and the patients have no or minimal co-morbidities. These are not the patients we take care of in the real world. Registry patients are typically much older, sicker and have multiple co-morbidities. We can harness data from our registries to inform, update and expand our guidelines.

Are there myths to dispel about registries?

There are two big myths. One is that registry data are of poor quality. The ACC has conducted audits and found that data are quite good—not at the level of a RCT, but certainly good enough to allow regulators and payers to feel confident to use the NCDR for post-approval studies, investigational device exemptions and post-market surveillance.

The other myth that troubles me is that registry risk-adjustment models aren’t good enough to take into account patients at very high risk—those with cardiogenic shock, for example. We’ve found that our risk-adjustment models work extremely well. We acknowledge that physicians have concerns, especially in a public reporting environment, so we are taking further steps to identify and properly categorize high-risk patients, such as out-of-hospital cardiac arrest, in the next iteration of the CathPCI registry. 

Can registries be mined to identify fraud and abuse, such as in the Department of Justice (DOJ) investigation of ICD implantation, where more than 450 hospitals settled with the DOJ?

This idea that the ICD registry was the genesis of that DOJ investigation is probably the biggest urban legend about registries. The fact is that the DOJ investigation started with a whistleblower lawsuit that was filed on Jan. 9, 2008, based on ICD implants performed prior to the 2006 creation of the ICD registry and three years before Sana Al-Khatib, MD, MHS, and colleagues published an article concluding that 22.5 percent of patients with ICDs in the registry did not meet evidence-based criteria (JAMA 2011;305[1]:43-49). This urban legend has taken on a life of its own! The NCDR ICD registry is actually a super resource for  hospitals and clinicians to use as a proactive resource for better assessing their indications and, therefore,  minimizing potential risk for accusations of fraud and abuse.

What are advantages of clinical vs. administrative data?

They both have incredible value. There is substantial concern about administrative data because it’s really billing data and lacking many of the assessments of clinical aspects of care. You can utilize both, especially for longitudinal follow-up of patients related to clinical outcomes. When you merge longitudinal administrative data with clinical data, you get a very powerful tool. For example, years ago, when there were concerns about the safety of drug-eluting stents (DES), Pam Douglas, MD, and her team at Duke very quickly, and at a cost of only about $100,000, merged three-year Medicare data with registry data for 262,700 patients to show that DES were safe across several subgroups (J Am Coll Cardiol 2009;53:1629-1641). We’re performing this type of analysis in the ICD registry to look at patients with borderline ejection fraction to identify subgroups of patients who benefit from ICD implantation, and similar analyses are being performed in the TVT registry.

With so many cardiovascular registries, how should administrators budget for staff to input data?

It’s not one-size-fits-all. It depends on what a provider’s specific needs are and which software vendor they utilize, among other things. The NCDR has free web-based tools for bare-bones entry, but many administrators are looking for tools that also meet other hospital needs, such as auto-populating data elements.

Data input by quality analysts is a substantial expense of doing business. Unfortunately, medicine is behind the times compared to other businesses. Every other business has invested in infrastructure for these types of analyses for assessment. As we move from volume to value, the cost will be easier for administrators to accept because it also will meet many other reporting needs. Where we’re headed, correctly, is the direct import of clinical data extracted from the EHR on to our data registries with the amount of direct, required data entry markedly descreasing. The PINNACLE Registry has an EHR-agnostic tool that is already doing this auto-population. It’s offering on-time, real-time value.