Registries Retooled For Better Data Mining

Simple to create and to extract data from, registries are becoming essential to effective patient and population management

Frank Diamond
Managing Editor

There’s a registry renaissance going on in health care. “Professional societies, providers, hospital associations all got together and now there’s this major push for the development of clinical registries,” says Joseph Cacchione, MD, chairman of operations and strategy at the Arnold & Sydell Miller Family Heart & Vascular Institute at the Cleveland Clinic.

Joseph Cacchione, MD

“Whenever you take two sets of data [clinical and claims] and merge them, there are logistical issues to resolve,” says the Cleveland Clinic’s Joseph Cacchione, MD.

We’re not just talking about lists of patients with the same disease. “Take a diabetes registry,” says Cacchione. “There are certain data elements that everybody collects in the same way for comparator reasons. How many of these people had hemoglobin HbA1c measured? What percent are on insulin and what percent are on an oral hypoglycemic? What different types of hypoglycemics are they on? There are specific clinical data points that are included in that registry.”

The Affordable Care Act encourages registries, points out Don Liss, MD, the senior medical director at Independence Blue Cross. “One of the requirements of the patient-centered medical home (PCMH) is to maintain registries. They can be fairly simple or they can be complex and require EHRs. It’s a fundamental change in practice for physicians.

“Most physicians are very comfortable being responsible for the patient in front of them on the exam table and running through the checklist of all the stuff that that patient needs at that particular moment,” Liss continues. “We’ve got to change the mindset so that that physician feels responsible for that same checklist, even if that patient isn’t in the office. The only way to get there is through registries.”

Insurer involvement in registries varies. Blue Cross Blue Shield of Michigan, for instance, takes a hands-off approach that it considers wildly successful. In 1997, BCBSM created Collaborative Quality Initiatives (CQIs), in which a consortium of providers supplies data in general clinical areas, such as surgery. BCBSM pays 80% of the cost of collecting and coalescing the data and also funds the coordinating centers where the data are stored for each CQI.

A policy initiative profile about the program on the Agency for Healthcare Research & Quality (AHRQ) Web site says, “Payments typically range from $10,000 to $125,000 per hospital per CQI, depending on the hospital’s case volume relative to the procedures being tracked in each CQI.”

The providers own the data, not BCBSM. About 70 hospitals contribute to 15 CQIs focusing on areas such as bariatric surgery, trauma care, breast cancer, and hospital medicine safety.

David Share, MD, MPH

“Registries just provide information,” says David Share, MD, MPH, of Blue Cross Blue Shield of Michigan. Who owns it and how it is harnessed are what really matters.

“These are clinically rich, comprehensive data,” says David Share, MD, MPH, senior associate medical director at BCBSM. “They include risk factors, processes of care, outcomes. It’s all contributed to a common registry and then the registry is used to examine the link between processes of care and outcomes using variations in practice across centers that participate in collaboratives.”


This gives providers a clear view of what works. “Members of the consortium systematically implement the new knowledge and they measure themselves using the registry. It’s a comprehensive and very rigorous approach to doing risk-adjusted comparison analysis,” says Share.

It’s not just scorekeeping, a clumsy measure of net result. “It’s using the registry in a very scientifically sophisticated way to judge how different approaches to care for different patient populations lead to different outcomes.”

The feedback from participating physicians and hospitals has been positive — at least 95% of network hospitals participate in all the CQIs they are eligible for, and 73% participate in at least one. “They are very excited because if they just look at their own data in their own institution, they don’t have enough information, nor do they have the analytic tools to learn much,” says Share. “When they pool the data they have a lot more data and a lot more variation, which is good for figuring out what works. And now they have peers across institutions who are eager to learn from one another.”

Follow the money

There are financial rewards. For instance, BCBSM contracts with hospitals contain a pay-for-performance element tied to CQI. “So the hospital, if it does really well and is fully engaged, can earn substantially more money,” says Share. “Recently we’ve begun implementing a physician-­recognition payment for the clinical champions from the hospitals to acknowledge their involvement in the project. They come to meetings, they’re on committees, they go back to their hospital, spread the word, and get committees engaged in trying to make change.”

The AHRQ says, “Because many hospitals participate in multiple initiatives, payments can range from $3,000 for a small hospital participating in only one CQI to $1.2 million annually for some large hospitals participating in multiple initiatives.”

The Cleveland Clinic, in partnership with HealthCore, a research company owned by WellPoint, is taking another route, bolstering clinical data with claims data. One HeathCore/Cleveland Clinic study of about 900 patients looks at bypass surgery; the other of about 3,000 patients looks at stenting.

“We should have preliminary results this fall,” says Cacchione. The studies, launched in January, have been challenging, he adds. “Whenever you take two sets of data and merge them, there are logistical issues to resolve. The data must be cleaned to allow them to be effectively merged.”

A combination of claims and clinical data should give providers a better idea of what is actually happening to the patient. “We look at five different clinical parameters. If they have two of those five, it predicts that their costs over the next two years are going to be X. If they have four out of five, their costs may be X plus Y. We tell the physician that when these two clinical parameters are involved, we need to take extra special care because these people typically are very resource-intensive. We’re trying to prevent unnecessary readmissions. Unnecessary tests. We can predict who’s going to use more resources.

“It’s about bundling. It’s about capitating. It’s about better understanding the individual patient and trying to create a situation where we can get more information in the provider’s hands so that we can more effectively manage these patients.”

For the most part, says Mark ­Cziraky, PharmD, HealthCore’s vice president, registries historically have had a limited frame of reference — typically from a single practitioner’s office (e.g., an endocrinologist for a diabetes study) — and miss the continuum of care, the visits to other specialists, pharmacy, and hospitalizations.

Mark Cziraky, PharmD

Mixing clinical and administrative data gives a much clearer view into what’s really happening with a patient, says Mark Cziraky, PharmD, vice president of HealthCore.

In other words, it’s typically been a snapshot, but Cziraky sees that changing to a more fluid view, almost like a movie.

“We identify a large group of people from our administrative claims system in a way that’s HIPAA-compliant,” says Cziraky. “We get regulatory approval to then identify the medical record to match the administrative claims — put those two together and analyze them in a de-identified fashion.”

Cziraky calls this a hybrid registry, including clinical and administrative data in a single design.

“When you start doing integrated data, you start getting much more clinical data elements that are not traditionally found in administrative claims data, such as family history, height and weight, smoking status, and other risk factors that allow us to conduct a more refined analysis,” says Cziraky. “A much clearer view into what’s really happening in that patient.”

Electronic health records might take some of the legwork out of building registries, says Cziraky — “as long as physicians are capturing the information needed for the registry. That’s a huge caveat, as there is a lot of inconsistency in data-capture by physicians within their EMR systems as part of their usual care.”

It might work this way: A patient goes to the endocrinologist. The endocrinologist will check the registry and see what the primary care physician did and what was prescribed. “EHRs are generally integrated based on provider group ownership,” says Cziraky. “My PCP and GYN both use EPIC but their systems aren’t connected, as the practices are owned by different groups. With just an EHR, neither of the docs can tell whether the patient has filled that script. That’s where claims data can help.”

Share asserts that the overriding driver of the success of BCBSM’s CQIs is professional pride.

“People, not just doctors, generally accomplish much more when they’re operating from intrinsic motivation,” says Share. “So if you just set up quality measures and hold them accountable, they’ll do what they have to do to earn their rewards, but they’re not going to do a lot more. But here they’re empowered with the potential for accessing improvement and they go for it. So there’s no question that ownership has dramatically increased what they’ve accomplished.

“We have ample evidence from many CQI programs of improvement in serious complication rates and mortality rates that outpace improvement seen in other parts of the country or, in some comparisons, in hospitals in Michigan which do and do not participate in a particular CQI program.”

An article in the April 2011 issue of Health Affairs that reviews the program states that “Michigan hospitals had more substantial improvements in rates of morbidity and mortality than other hospitals participating in national data feedback programs administered by the Society of Thoracic Surgeons and the American Collage of Surgeons.”

The Health Affairs article says that the program “reduces payments associated with adverse outcomes by approximately $20 million annually — far exceeding the $5 million annual cost of administering the program.”

If BCBSM owned the data and decided who is producing and who isn’t, providers might dismiss the findings.

“Legitimately so, because no data are perfect,” says Share. “When they own it, they are quite willing to take responsibility for variation in practice outcomes and to use that information to catalyze improvements, more so than if someone from the outside were making judgments.”

Insurer’s role

BCBSM plays an important role, nonetheless. “For one thing, there is nobody else who can convene otherwise competing physicians to come together and collaborate,” says Share. “If one of them just said, ‘Why don’t we get together and do the registry in the way that I’m describing’, the others would say, ‘You probably want that for your advantage and we’re not going to play.’ But if the Blues say that it’s really important for the provider community to up the level of the use of scientific tools to optimize care, they’re going to support that.”

Avoid duplication

There are about 120 clinical registries, says Joseph Cacchione, MD, chairman of operations and strategy at the Arnold & Sydell Miller Family Heart & Vascular Institute at Cleveland Clinic. The American Medical Association recently launched an initiative to try to “harmonize some of these registries so that you don’t have three different registries all looking at the same topic,” says Cacchione. “We don’t want competing registries because so many of these things cross different specialties.

“For example, the vascular surgery registry looks at putting stents in legs. Well, vascular surgeons aren’t the only people who do that. Cardiologists do it. Radiologists do it. So they’re looking to try to harmonize some of this information. Organizations like the AMA are looking at these registries to make sure they are measuring the same things in the same manner.”

Registries compare hospital performance, says Cacchione. For instance, the Consumers Union Web site posts a Society of Thoracic Surgeons registry. About 10% of hospitals are three-star programs, according to the society and Consumers Union, says Cacchione. “About 70% are two-star and 20% are one-star. If your hospital is a two-star program, you look at what a three-star program does and make the determination to be a three-star program. These are the things that drive performance improvement.”

A blueprint for high-volume, high-quality lung cancer screening that is detecting cancer earlier—and helping to save lives

Clinical Brief

Multiple Sclerosis: New Perspectives on the Patient Journey–2019 Update
Summary of an Actuarial Analysis and Report