By 2030, cardiovascular disease will affect 40% of the population of the United States, according to projections by the American Heart Association.
Taking this into consideration, hospitals and other healthcare organizations will need a data-driven plan to treat those heart patients and to manage costs.
One organization has already carried out a successful plan. Allina Health, based in Minnesota, worked to focus on cardiovascular (CV) care in its system of 12 hospitals, 65 clinics, 49 rehabilitation locations and 23 hospital-based clinics.
By making data-driven decisions, testing new processes, examining population health issues and clinical variations, the organization saved more than $155 million and saw $46 million bottom-line improvement.
In the past three years, they’ve avoided:
- 1,017 ICU admissions
- 142,194 lab tests
- 3,220 days in the hospital
At the 2018 HIMSS conference in Las Vegas, Allina Health CV Clinical Program Director Pam Rush, RN, MS and Dr. Craig Strauss, a cardiologist at the Minneapolis Heart Institute, profiled Allina’s strategies and how they were successful.
“We’re not getting the ROI that we’d like to see on healthcare expenditures. American life expectancy does not match the level of spending, and 17% of national health expenditures are spent on cardiovascular disease,” Rush said. “But what does that mean to consumers? It means our premiums, our deductibles, our co-pays are going up. Hospitals that can reduce cost, those are the ones that are going to survive. Those that can’t will have a much harder time.”
Becoming a Data-Driven Organization
Allina Health had to develop an enterprise data warehouse to look at the data in different ways. Its data scientists obtained that data from different sources, so they could break it down by different variables. For example, they examined data by hospital and procedure types, giving them varying ways to check out the data and make improvements.
“Understanding the impact of our care better, understanding the risk for the patient, will allow us to better negotiate on behalf of our patients. We need to move toward more risk-based models,” Rush said.
They used analytics tools and resources, as well as clinical involvement. They also developed dashboards to help look at data and interpret it.
“We have a pretty tight data governance infrastructure,” Strauss said. “We have our healthcare delivery innovation team who can control who sees the data, and we’re relatively cautious with how much we share the data. We want to make sure it’s used in the right way. Going forward, it’s about identifying areas where clinicians feel like they’re getting real-time data that’s meaningful and gives real-time characteristics of patient populations.”
Improving Outcomes and Decreasing Costs
Strauss said that when they address patient characteristics such as gender and age, they can then determine the risk of internal bleeding and try to avoid it in patients undergoing percutaneous coronary interventions. By making informed decisions, they can assess the available options, from devices to pharmacological solutions.
“We’ve seen a correlation between bleeding risk and length of stay,” Strauss said. “High bleeding risk is a key metric, as … patients (with excess bleeding) suffer from complications and higher mortality rates. So, you want to be able to know this information before a procedure. We took this to our interventional cardiologist and told them that not only can they identify high-bleeding-risk patients, you can do something about it.”
In 2012, Allina Health noted a disparity between physicians using closure devices and those who weren’t. Four years later, about 78% of procedures performed on high-bleeding-risk patients were performed using the closure devices.
“We had to effectively communicate the risk to providers so they could communicate with people in the lab and make the choice as to whether or not they’d use the closures,” Strauss said.
Another big factor in CV outcomes is post-operative atrial fibrillation. About 30% of their patients suffered from it, within national statistical norms, and these patients were costing around $7,000 per day more than other patients.
Based on this, they developed a standard protocol for nurses to follow so they didn’t have to consult with different physicians each time – they just followed the protocol. After implementing the protocol, they found that the average patient length of stay decreased by about two days.
“We looked at all the patients we had with post-op AF, because if you want to manage this population, you have to identify all the cases, then focus in on the ones you have the most granular data and best understanding of,” Strauss said.
They also performed zip code analysis on specific patient populations to determine whether “care was comparable across the continuum,” Strauss said. Patients who have an acute artery blockage are taken directly to the catheter lab, whether they came from the emergency room or were transferred from rural areas.
“What we found is that there was similar care for patients who were transferred in, and in many cases similar outcomes,” Strauss said.
Addressing Questions and Concerns
At HIMSS, Strauss participated in a brief Q&A session to answer further questions from the audience.
How widely is data-provider-related data shared? How do you approach that issue of transparency in sharing provided related information?
“We’ve wrestled with that to be honest,” Strauss said. “Doctors often wanted to know everything and wanted to see how their partners were doing. We shared everything down to the individual physician with everyone. But there are other areas where we don’t share quite as transparently, because it can be used in negative ways. You may have two hospitals competing for resources and they’ll want to use that data in ways that support those resource initiatives, rather than focusing on using the data to improve quality.”
What do you see as the next frontiers of data? When we think about CV health, in particular, there are social determinants of health (SDoH) factors and genomic factors, so as you start to think about context and additional data sources, where do you think those new data sources will be most impactful and how do you anticipate incorporating those data sources into your work?
“One of the areas we’ve begun to delve into, but we don’t have great data, is our payer information. We have data on overall cost of care from certain populations, but I think there is an opportunity to bring in all kinds of payer data and link it with medical data. We’re also developing an app for patients who see us [at the] clinic to download. We’re hoping to have patient-recorded outcome data back into the app and the app downloads that into our data warehouse, so we can link patient recorded outcome measures to all the other clinical data we have. Across the country, we don’t do enough to understand what patients are feeling and what their quality of life is.
“One of the other things we’re doing is looking at gaps in care. People often joke that your credit card company can tell you when you’re going to have a baby. We aren’t doing enough to proactively reach out to patients who we know are at risk or are candidates for certain interventions. We’ve been working to identify gaps in care and opportunities to reach out to patients.”
What do you see as the role of the life sciences industry in this movement toward quality and data analytics driven healthcare? Do you see pharma and med device companies as being a partner or a distractor in this?
“I think partner. Particularly when I think about med device companies and cardiology, I think about pacemakers and defibrillators that collect tremendous amounts of data we don’t often link back to our own clinical data. One example is that many pacemakers can identify patients who have transient atrial fibrillation. We don’t link that back effectively to identify patients who should be on blood thinners to reduce the risk of stroke depending on the frequency of their atrial fibrillation and what other characteristics they have. I think there’s a real partnership opportunity.
“We have to be careful, because there is so much data and we have to be able to pick what the right data is and use data somewhat sparingly so we don’t overwhelm people. When you think of how much data is available from insurance companies, device manufacturers, etc., it’s important to filter what’s meaningful and not overwhelm the clinician.”