Yale-New Haven Health System’s Bold Plunge Into Analytics-Driven Performance Improvement

Jan. 7, 2015
Thomas Balcezak, M.D. and Stephen Allegretto are helping to lead a long-term initiative at Yale-New Haven Health System that is using analytics to both standardize clinical processes and reduce care delivery costs organization-wide

At a time when virtually all patient care organizations in the United States are under pressure to simultaneously improve clinical outcomes and reduce patient care costs, figuring out successful strategies and pursuing practical tactics remain major challenges for most. At the Yale-New Haven Health System in Connecticut, finance and medical leaders are pursuing a promising strategy that is profoundly data-driven.

There, Thomas Balcezak, M.D., senior vice president and chief medical officer, and Stephen Allegretto, vice president of strategic analytics and financial planning, are helping to lead an exciting initiative that is beginning to bear fruit, at the health system, which encompasses three hospitals (the 1,500-bed Yale-New Haven Hospital in New Haven; the 350-bed Bridgeport Hospital in Bridgeport; and the 100-bed Greenwich Hospital, in Greenwich). Yale-New Haven Health System is a $3.3 billion health system in annual revenues ($2.4 billion of which are derived from flagship Yale-New Haven Hospital), with a medical staff of 4,000 physicians and 12,000 employees.

Dr. Balcezak and Allegretto are spearheading an initiative in which they’ve created an ongoing analytics process focused on what they’re calling quality variation indicators (QVIs), measures that are helping YNHHS leaders link clinical, financial, and operational factors together to determine where process improvements can be made. On Oct. 21, 2014, the two executives presented on the topic “Engaging Physicians to Drive Financial Performance,” at the Strata Decision Executive Leadership Symposium, held at the Swissotel in downtown Chicago and sponsored by the Chicago-based Strata Decision Technology, which is a vendor partner in the ongoing Yale-New Haven Health Care initiative. Dr. Balcezak and Allegretto spoke recently with HCI Editor-in-Chief Mark Hagland regarding their ongoing work in this important area. Below are excerpts from that interview.

You and your colleagues are using the phrase “cost of quality” to describe some of what you’ve been working to analyze.

Stephen Allegretto: Yes, one of the tenets we’ve been working with is identifying quality variation in a manner that both physicians and nurses could identify with. So we came up with the concept of “quality variation indicators,” or QVIs, phenomena we’ve identified that happen to patients in the hospital that we don’t want to have happen to patients in the hospital. About seven percent of our patients, out of 80,000 discharges, experience these QVIs. A QVI s a comprehensive definition of quality and safety; these patients stay on average about three times longer and cost about four times as much. So by looking at our patients through that lens, our clinical folks have identified that yes, these are the patients we care for, as they’re the same patients they’re evaluating for morbidity and mortality, or as we evaluate our healthcare-acquired conditions data…

What are some of the key indicators you’re looking at?

Thomas Balcezak, M.D.: Our QVI work really is a summarization of a lot of the stuff that’s been in the medical literature for years. There are estimates about what a hospital-acquired bloodstream infection costs, for example, anywhere between $20,000 and $60,000. So there are a number of different characterizations of these types of events. Patient safety indicators, PSIs [patient safety indicators], HAIs [hospital-acquired infections], and others, these types of measures are all on DRG- and ICD-9-driven lists. And Steve and I and our colleagues looked at a relevant list of PSIs and other commonly coded conditions and worked through what was thought to be possibly preventable and what was thought to e clinically relevant, and created the QVI list. It’s an amalgamation, with some clinical expertise overlay.

And it’s resonated with us because it’s been driven by clinical expertise on what’s possibly preventable and clinically relevant; and second, when we did tests about whether these QVIs made sense, we looked at two sets of data. One was financial data. One factor was that only 7 percent of our discharges accounted for 28 percent of our costs, where month over month and year over year, the amount and proportion of those patients really held steady over time. And that became even more interesting when we looked at some of our clinical data bases and registries. The clinicians really resonate with the registries, because they come out of specialty and subspecialty societies. And when Steve’s team lined up the registry work with the coded medical work and QVI work, we were able to obtain the Rosetta Stone of medical language, because it brought together the financial and clinical data.

What are a few examples of the most typical QVIs?

Three of the key ones we’ve run data on and analyzed so far have been accidental puncture or laceration during a procedure; postoperative deep venous thrombus and pulmonary embolism, or blood clot; and post-procedural pneumothorax, or punctured lung. They’re not talked about as much as catheter-associated bloodstream infection, which is on the list, too. Interestingly, in every instance with those three QVIs, we found that our medical record coded data and registry data were concordant; what’s more, in two of the three, there was a clinical practice issue that we needed to address, and in one of the three, a coding issue. Example, let’s talk about DVT [deep venous thrombosis] prophylaxis. There’s a wide disagreement in the literature about what is needed after a hip or knee replacement to prevent DVT. But here’s the issue from my perspective as a clinician: standardization in and of itself, is still worthwhile.

Why is it worthwhile with regard to the treatment of DVT prophylaxis, for example?

It’s worthwhile for a couple of reasons. One is that if we allow 17 different ways for post-hip and post-knee surgery DVT prophylaxis to be performed, and then we go back and look at whether DVT was applied in a timely way or correctly, we then need to look at 17 different sets of practices. Second, if we’re going to routinize this and take DVT prophylaxis out of the hands of the physician and into the hands of a pharmacist, we can’t teach the pharmacist 17 different strategies, only one.

Allegretto: And when a QVI stretched across interventional cardiology and orthopedics, we also looked at variation across non-QVI cases.

In the absence of extremely strong consensus in the clinical evidence, is it difficult to achieve standardization consensus among physicians?

Allegretto: We know there’s great variation in outcomes. What causes it? Asking the question has helped move that discussion forward, just by us being able to provide the data; as Tom said, there’s now high concordance in our data from the coded data to the clinical registry side. So they know there’s that concordance in that area, and they’re recognizing that maybe standardization will address some of that variation.

Balcezak: Not because one approach is better than another, but standardization just for operational reasons is an appropriate reason to standardize. So if you can’t tell me this regimen is superior to the other; then we need to have a discussion about the operational benefit of standardization of regimen/process. And the fact is, our presentation resonates with every audience we speak to. We gave a similar presentation at UHC [University HealthSystem Consortium, an alliance of academic medical centers] after the Strata Decision one. And we had people from an organization we really admire, who came up to us and said, ‘Holy smokes, you guys are way ahead of us on this!’ And that surprised us.

And we’ve got the tools now—analytics, and our registry resources, and everything else—we’re now able to do things that we were not able to do. And it points up some very helpful questions to help drive us towards things we wanted to accomplish, like around standardization, and the tools have helped us.

Are physicians ready to agree that unnecessary variation is undesirable in the healthcare system?

Allegretto: I don’t believe there was an agreement on what the variation was, generally speaking. And now we’re able to show that there is variation that we’ve been able to document, on the quality side. And now we’ve also been able to document variation in waste, on the non-quality side. And I believe physicians were aware of this, but there wasn’t the data. And Tom and I have been around this for a long time. And I believe they acknowledge that we do too many MRIs, too many lab tests, and through our EMR system, we’re able to deliver some of that data at the time they’re ordering, so the discussion is a bit different now.

Balcezak: Yes. You asked to understand the mindset of our medical staff. There’s a helpful way to think about the mindset of the medical staff, in terms of politics. Are voters accepting of gay marriage, or not? That’s changed a lot among the electorate in the last decade, hasn’t it? But then there’s also the overall direction of thought. If you used that analogy across our medical staff, you could draw some comparisons. There are members of our medical staff who are extremely progressive in their thinking around the cost of care or standardization, and some who I believe have antiquated thinking. So generational change among physicians, our ability to deliver to them good data with excellent comparisons, using UCH and other benchmarks, the overlay of registries, all those are factors. But if you speak of individual medical staff members, they’re not that different from 10 years ago. But the direction of thought is changing.

Allegretto: And Tom and I were the only team of presenters drawn from both finance and medical staff at the UHC conference. That’s changing. And another physician, Ryan O’Connell, presented with me at the IHI [the Cambridge, Mass.-based Institute for Healthcare Improvement] Conference in December. It was mostly physicians in the room; and Ryan asked the audience, how many of you have done work with finance people and have had tension with the finance people? And nearly everyone raised their hands in agreement. But we have now been able to cross over into collaboration with medical staff. And I think with that partnership between finance and medical staff will be key going forward.

And you need data at the core, correct?

Allegretto: Absolutely. We have good data from registries, the EMR, coding, etc.

Can you share any high-level metrics around your current progress in this work?

Balcezak: With regard to improved productivity, we’ve actually reduced the cost per unit of service across all three organizations, based on a combination of factors. Across all three organizations, we know that just in the last year, we’ve reduced our cost per unit of service at Yale-New Haven Hospital by 1.6 percent; at Bridgeport Hospital, by 1.8 percent; and at Greenwich Hospital, by 3.5 percent. It’s important to note that those reductions in cost are taking place at a time when other patient care organizations have seen their costs per unit of service go up. Still, at the same time, our revenue per unit of service has declined at the same time that we’ve reduced our cost per unit of service, and we know that the revenue per unit of service will continue to decline over time, so we need to decrease variation. In fact, this is the first year in which our revenue per unit of service went down across all service lines, but we know that that trend will continue in the future. So we’re moving forward among a number of quality and service line teams focusing on those QVIs going forward. We want to move that needle to improve our operating margins.

Allegretto: At the IHI presentation, we also asked folks in the audience how many people knew what their revenue per unit of service and their cost per unit of service were, and fewer than 3 percent could raise their hands. So we’re ahead of the industry in that way.

Balcezak: And per what Steve said about Ryan O’Connell asking about teaming between finance and medicine, can you imagine the people making iPhones not knowing what their cost per unit of manufacturing iPhones was? Or knowing the cost per unit of service? And Steve and I were taught finance management by the same woman, and if our inputs are drugs and surgery, etc., and our outputs are discharges etc., how can you not pursue cost per unit of service? And 15 years ago, Steve began building a system at Yale-New Haven to be able to do this.

What will happen in the next couple of years at Yale-New Haven Health System?

Allegretto: We’ll continue to make the sausage better as we go forward; we know we’ll get paid less per unit of service, and we’ll need to be able to manage not only our costs, but those patients we take risk for.

Balcezak: I’d say the same things, underscoring that we’re going to be able need to e able to manage episodes of care, not just intervals of care. And the length between cost and quality will be a big area for us to work on, because there’s a lot of opportunity to reduce cost and improve quality. And no matter how the world evolves, no matter what changes happen in healthcare, that’s a strategy that can’t go wrong. To continue to build on the foundation we established, will continue to be the essence of our strategy.

Sponsored Recommendations

Care Access Made Easy: A Guide to Digital Self-Service for MEDITECH Hospitals

Today’s consumers expect access to digital self-service capabilities at multiple points during their journey to accessing care. While oftentimes organizations view digital transformatio...

Going Beyond the Smart Room: Empowering Nursing & Clinical Staff with Ambient Technology, Observation, and Documentation

Discover how ambient AI technology is revolutionizing nursing workflows and empowering clinical staff at scale. Learn about how Orlando Health implemented innovative strategies...

Enabling efficiencies in patient care and healthcare operations

Labor shortages. Burnout. Gaps in access to care. The healthcare industry has rising patient, caregiver and stakeholder expectations around customer experiences, increasing the...

Findings on the Healthcare Industry’s Lag to Adopt Technologies to Improve Data Management and Patient Care

Join us for this April 30th webinar to learn about 2024's State of the Market Report: New Challenges in Health Data Management.