The 2018 Healthcare Analytics Summit: Wednesday Recap

September 13, 2018

Article Summary

The first full-day of the 2018 Healthcare Analytics Summit (HAS 18) featured keynotes from Marc Randolph (Co-Founder, Netflix), Dr. Brent James, Dr. Daniel Kraft, Dr. Toby Cosgrove, Dr. Jill Hoggard Green, and Dr. Robert Wachter. Two waves of breakout sessions covered success stories from organizations all over the nations, complete with countless lessons learned.

HAS 18 Logo

Focusing on digital innovation and disruption in healthcare, the Healthcare Analytics Summit™ (HAS) 2018 kicked off with a welcome from Health Catalyst COO Paul Horstmeier, who also introduced the new and improved HAS app. The app gives details about sessions, allows for connections with speakers, provides networking opportunities with peers, and of course, is the platform for playing (and winning) the HAS game.

Dan Burton, CEO of Health Catalyst, opened the HAS 18 general session announcing that 2018 marks the summit’s fifth year and is its largest to date, with about 1,500 attendees. Mr. Burton announced two days of in-depth discussion of digital innovation and disruption in healthcare. He explained that keynote speakers and breakout session presenters would focus on current digital transformation in healthcare and look to learn from transformation experiences in other industries (e.g., What does it take to disrupt or be disrupted? Will you be a Netflix or a Blockbuster?). Transformation topics, said Mr. Burton, would cover organizational and cultural perspectives as well as innovation in the employer and payer spaces and in wearable technology (e.g., biosensors) and price transparency. A key theme across presentations would be how do we do more with less, as the pressure to embrace digital innovation escalates in the face of rising healthcare cost restraints. Mr. Burton asked the HAS community if they’d face the transformation challenge with known tactics or take a risk and try something new. He promised all would come away with tangible ways to transform. Our shared challenge is that we must transform to survive and thrive, but the transition to value-based care requires doing more with less. How can we embrace digital innovation and disruption while still operating within the current atmosphere? Our commitment during HAS 18 is to provide specific real-world examples of transformation.

Keynote: Why Clinical Quality Should Be Your Core Business Strategy

Brent C. James, MD, MStat
Former Vice President and Chief Quality Officer Intermountain Healthcare

According to Brent James, MD, MStat, former Vice President and Chief Quality Officer at Intermountain Healthcare Transformation, healthcare is poised for a transformation not seen since internal medicine revolutionized healthcare more than a century ago. Like the modern medical pioneers of the early 1900s, today’s healthcare professionals (from data scientists and finance experts to executives and frontline clinicians) have immense opportunity to create a new, better direction for healthcare, in which care improves and costs drop.

Dr. James bases his outlook on his work at Intermountain in clinical variation. Focusing on transurethral resection of the prostate (TURP), he applied clinical research measurement to care delivery performance. He found massive variation in TURP surgery time and amount of tissue removed—even among high-volume surgeons. Data on treatment failure (men who re-obstructed within one year of surgery) helped him determine the true cost to the hospital of TURP surgery variation and identify a level of variation beyond quality care for all as well as unacceptable rates of preventable care-associated patient injury and death. Dr. James recognized a significant opportunity for improvement by targeting waste (up to 50 percent of healthcare resources) in care process that lead to spiraling prices, which limited access to care expenditure is waste. “The future of healthcare lies in waste,” he said. By targeting waste from the supply chain to utilization levels and aligning clinical interest in the form waste reduction with financial incentive, Dr. James decreased clinical expenditures at Intermountain by 13 percent 2016. The lessons, he said, are to count successes in lives saved, operate under the philosophy that better care is cheaper care, and measure success against good internal data. As pay for value is the future and will continue to grow, Dr. James explained, health systems must prioritize clinical quality and align financial incentives accordingly.

In 2018, the quality of patient care is falling short of its potential. Dr. Brent James shared five reasons for this:

  • Massive variation in clinical practices.
  • High rates of inappropriate care.
  • Unacceptable rates of preventable care-associated patient injury and death.
  • Inability to do what we know works.
  • Huge amounts of waste, leading to spiraling prices that limit access to care.

With these challenges comes massive opportunity. If healthcare providers can achieve miracles in care delivery for patients and manage to do it correctly half the time, what could we accomplish if we manage to get it right close to 100%?

Organizations have much higher financial leverage from waste elimination than revenue growth. Dr. James shared the shocking statistic that between 30 and 50% of all healthcare resource expenditures are quality associated waste. This waste results from preventable screw ups, building unusable products, providing unnecessary treatments and simple inefficiency. Thus, reducing waste is necessary to survive financially.

In the drive to create a new generation of care delivery, Dr. James shared four lessons:

  1. We count our successes in lives.
  2. Nearly always (with proper clinical management) better care is cheaper care through waste elimination.
  3. None of our successes came from comparing ourselves to others – it comes through solid analytics based on good internal data.
  4. The long-term organizational viability of clinical quality improvement strategies requires aligned financial incentives.

He also speculated about what the future of healthcare might hold:

  • Pay for value will continue to grow
  • Health IT will mature
  • A fundamental shift in focus through care delivery will take place.

Keynote Presentation

Daniel Kraft, MD
Faculty, Chair for Medicine & Neuroscience, Singularity University, and Founder, Chair, Exponential Medicine

Dr. Daniel Kraft, founder of Exponential Medicine, provided a fast and furious ride through the “uberization” of healthcare. Much of medicine is still focused on sick care that’s episodic and reactive, rather than focusing on health care that is continuous and proactive. Better, cheaper, and faster is the future of true healthcare. We need to move the needle in that direction, not only to save money but to improve wellness and add years to lives.

We’ve created a massive wave of siloed healthcare data. The future of healthcare needs to connect those dots with data and make it actionable. Dr. Kraft used examples such as fitbits and fitness apps that contain so much data about not only health, but also behaviors. He said that behaviors, more than genetics drive healthcare costs.

What could data contained in these new fitness apps tell your doctor about how to improve your health? He explored fascinating questions about the ability to crowdsource and democratize data in order to drive true change.

Dr. Kraft discussed innovative ways healthcare can leverage technologies to take the needed exponential steps towards digitalization. Here are just several of the advances he mentioned:

  • Wearable devices will not only measure but eventually diagnose treat health conditions.
  • Innovators will find ways to use digital output (e.g., social media) to measure mental health status and leverage that data for behavior change.
  • Digital platforms will help democratize medical education (e.g., surgeon will use virtual reality to practice procedures).
  • Genome sequencing will help identify which drugs and dosages are optimal for each patient, predict disease, and more.

However, having data is not enough. It’s all about making the data useful. How do we leverage data into behavior change?  No one wants another checkbox or another 12 clicks to prescribe medication. We need to go beyond the EMR to truly patient centered care. While there are so many innovations in technology and healthcare, Dr. Kraft said it’s not doctor vs. machine, but how we are able to integrate them.

Keynote Session

Toby Cosgrove, MD
Former President and CEO, Cleveland Clinic

HAS 18 attendees were treated to a Q and A session with healthcare giant Toby Cosgrove, MD. As a former veteran, cardiac surgeon, and CEO of Cleveland Clinic, Dr. Cosgrove’s list accomplishments is too long to count. Dale Sanders, President of Technology at Health Catalyst, steered the discussion with questions from “Is Google going to own hospitals?” to “How do you train empathy?” Attendees were treated to Dr. Cosgrove’s insights on the future of the VA, the role of data and digitization in the recruitment and retention of physicians, and how he was able to overcome his dyslexia and go on to such a multitude of achievements. The conversation touched on Dr. Cosgrove’s personal experiences and how they inform his perspective on pivotal topics in healthcare today:

Leadership in the Data-Driven Healthcare Era

Healthcare organizations need ways to handle growing healthcare data (to store, manage, learn from vast amount information).

Clinical Trials Versus Data

Research could better understand clinical trial results better if they collected data they as moved forward. Dr. Cosgrove speculated the data could even replace clinical trials.

The Role of Google, Amazon, and Other Major Technology and Commerce Players in Healthcare

These organizations have great capabilities that can add tremendous knowledge and insight to healthcare organizations (e.g., Microsoft’s HoloLens, which can teach anatomy without cadaver) and superior cybersecurity. Healthcare has much to gain by partnering with other industries to solve problems.

“Patients First”

Leadership must take responsibility for healthcare and know who they’re responsible for—the patient. Dr. Cosgrove’s “patients first” buttons drove home this point, and he used the philosophy to reorganize care around the patient’s needs.

Data Is Non-Negotiable, but Needs Better Management

Health systems must have data capability now. Healthcare, however, is overwhelmed with data, and changing the clinician-patient relationship and contributing to clinician burnout, as clinicians are burdened with EMR data entry.

Keynote Session

Marc Randolph
Co-Founder, Netflex

HAS 18 speakers and attendees have been using the phrase, “be a Netflix, not a Blockbuster” to refer to the two paths in healthcare digitization: be the disruptor (Netflix) or the disrupted (Blockbuster). Marc Randolph, Netflix Co-Founder, Executive Mentor, and Angel Investor shared his path to disruption with Netflix and why it’s a thriving enterprise today, while the previous leader in home movie rentals, Blockbuster, went bankrupt in 2010.

Marc began by admitting what he said was an embarrassing fact: his favorite place in the world is Dallas, TX. The reason? It’s the location of the Renaissance Tower–and the headquarters of Blockbuster Corporation. In 2000, Netflix was 2.5 years old, had less than 100 employees, and were on track to do about 3 million in revenue. They had also spent about $50 million to get there. In other words, it was not a sustainable model and they were looking to sell a portion of the business to Blockbuster.

At the time, Blockbuster had 60,000 employees and were on track to do six billion dollars in sales that year. After several attempts, they got a meeting in Dallas and were excited and hopeful about the outcome. Then they were asked a number: “How much should we pay for you?” The answer, $50 million, was met by silence and the meeting quickly went downhill.

Marc was forced into the realization: we’re going to have to take them down. This story about taking down the market leader is inspiring for innovators, but it’s a different story if you’re the market leader. If you want to avoid being disrupted, you have to figure out how to disrupt yourself.

Mr. Randolph shared lessons from his experience of taking down a $6 million market leader with a startup that had $50 million in debt:

  • Innovators don’t need to be Silicon Valley to innovate. In the digital era, innovation is happening across the globe.
  • Innovators don’t need special training. Some disruptors have little background in the field they disrupt. They just see a need (a pain point) and address it.
  • Innovators don’t need to be the best and brightest.
  • Innovators need a tolerance for the risk of not knowing what’s next.
  • Innovators need the capacity to generate ideas (hundreds or them), but the ideas don’t need to be big or even original. The important part is to have the idea.
  • Innovators need confidence to push for right solution. No one knows if an idea is good or bad until they test it.

Since leaving Netflix, Marc has had the opportunity to lead many early stage companies and he’s learned a few important lessons:

  1. You don’t need to be in Silicon Valley.
  2. You don’t need special training.
  3. You don’t need to be the best and the brightest.

So if those things aren’t important, what is important?

  1. Tolerance for risk.
  2. The capacity to generate ideas. The ideas don’t all need to be big and they don’t all have to be original. They don’t even have to be good ideas (because what is a good idea?)
  3. Confidence. When people tell you your idea is not going to work, most of the time they’re going to be right. You have to have the confidence to try anyway.

Breakout Sessions: Wave One

Session 6: How We Developed an Advanced Analytics Team to Solve Our Strategic Problems

David Wild, MD
Vice President, Performance Improvement, The University of Kansas Health System

Chris Harper, MBA, MPM
Vice President, Applications and System Development, The University of Kansas Health System

Before starting their improvement journey, the University of Kansas Health System was in trouble. The system had low patient satisfaction, high mortality rates, and more medical students than patients. Twenty years later, it’s a 925-bed, $3 billion organization that ranks high inpatient satisfaction with low mortality rates. It achieved this success by focusing on making data-driven improvements.

Starting in 2013, The University of Kansas Health System began the data journey:

  1. Recognize the need to change.
  2. Conduct a current state assessment.
  3. Identified a solution
  4. Implemented a phased approach

When Chris Harper did his initial analysis, he found that disparate data sources was a consistent problem across the organization. He also found that the data analysts were spending 60% of their time answering questions and creating reports, and 31% of their time hunting and gathering data. They needed an integrated approach.

To get buy-in from leaders, Dr. Wild presented a pitch that presented the following idea: we regularly make decisions based on available data, but we rarely know what could have been if different data was available. What they wanted to do is take data and, as quickly as possible, have analysts develop that information and translate it into actionable knowledge that can lead to outcomes improvement. If they could do that better and fast than competition, they would have better results. They were given the go ahead to assemble a team that could prove they should move ahead with this journey.

He shared the following lessons and recommendations as you build your own road map:

  1. Think it through.
  2. Do it your own way.
  3. Don’t forget the why.

Session 7: Using Machine Learning and Big Data to Drive Patient Engagement and Better Health Outcomes

Alexander (Alex) Marano
Customer Analytics Lead, Cigna Information Management and Analytics

Christer A. Johnson
Principal, EY Analytics

Acknowledging that customer engagement is a critical step in better outcomes, Christer A. Johnson and Alexander (Alex) Marano presented on using machine learning techniques to analyze a broad range of patient data. The key to improving engagement, they explained, is to leverage health information to identify the most impactable moments in a patient’s journey and plan to engage at those moments.

As the breadth of and access to health data grows, the industry can understand more about the customer journey; whereas health systems previously relied largely on claims data, they now have time-sequenced journey data steps (demographic, lab, call center, and click-stream data from web interactions and mobile phone interactions) to create a fuller picture of the patient healthcare journey. This comprehensive view helps them understand who’s engaged and the patterns most associated with engagement, as well as when to best intervene to increase engagement.

In a pilot of health coaching for case management populations, the presenters identified key journey steps that impacted engagement (e.g., reaching out within a week of an office visit), applied journey analytics, and compared engagement with and without health journey analytics. This helped them understand likelihood to engage and find opportunities to increase engagement. Overall, they found opportunities that increased engagement by 30 percent.

Session 8: Real-World Examples of Leveraging NLP, Big Data, and Data Science to Improve Population Health and Individual Care Outcomes

Shaun Grannis, MD, MS, FAAFP, FACMI
Director Regenstrief Center for Biomedical Informatics; Assoc Prof, Dept of Family Medicine, IS School of Medicine

Dr. Shaun Grannis began his presentation with the simple premise that caregivers aided by thoughtful information technology are more effective than caregivers alone. Machine learning has tremendous potential to improve health outcomes. But even with the exponential growth in the volume of health data we have today, we still need more and better data and tools to improve the efficacy of our machine learning models. To that end, Dr. Grannis shared a few essential ingredients, for example:

  • Health information exchange. To be successful, we need to be able to take data from multiple sources and multiple providers—and then standardize, integrate, connect, and automate it.
  • Text analytics/natural language processing. The structured data we collect from claims and electronic health records only represents about 20% of available information. We need to improve our mining of free text sources such as chief complaints, nursing notes, and surgical pathology reports.

Dr. Grannis showed that while machine learning tools are in their infancy, there are already examples of real-world success:

  • High utilization prediction from ED surveillance data. How emergency department event surveillance data was used to discover and target the biggest predictors of high utilization: chronic disease coupled with a mental health condition.
  • Cancer identification from free text. How cancer patients were identified from sensitivity analysis using random forest applied to positive cancer mentions from surgical pathology reports.
  • Targeting of wrap-around services. How a community care prediction model identified patients most in need of referral to nutrition, financial, and social resources.
  • Notifiable disease detection. How automated machine learning models were shown to be far superior than humans to identify and report notifiable diseases.

Session 9: Real Quality: A Recipe for Healthier Patients and Happier Doctors

Christian Dankers, MD, MBA
Associate Chief Quality Officer, Partners Healthcare; Harvard Medical School Faculty

Dr. Christian Dankers started the session with the Triple Aim, Plus One—Partners HealthCare strives to provide the best care for the patient, across the population, in the most cost-efficient way. The “Plus One” is Clinician Satisfaction—and all the other elements of the Triple Aim depend on the satisfaction of the clinician.

Partners sought a better way to think about measures, with the goal of improving care. As they began this work, they discovered that it was a journey:

  • More clinically relevant measures—measures needed to be unifying, motivating, accurate, available, actionable, and balanced and parsimonious; they needed to be more diverse (e.g., include specialty care)
  • Increased buy-in from clinicians—predictive modeling helped determine whether efforts yielded benefits (e.g., by measuring risk reduction); additional efforts include safety and PROMs measures
  • Increased investment in tools and effort; improve on clinically relevant measures—leadership engagement, transparency, policy and incentives, training and education, EHR and electronic tools, and process improvement
  • The result: better care. Partners has seen successful quality improvement in clinical areas such as cervical and breast cancer screening, lipid control, BP control, and others.

Dr. Dankers closed by saying the Partners vision is to spread clinical measures across the continuum of care. The vision of clinically relevant metrics that represent true care is possible by engaging clinicians, giving them the tools they need, and providing the support that allows them to improve—this results in better care for patients.

Session 10: Integrating Data and Analytics into Provider Workflows Improves ACO Quality and Financial Performance

Joan Valentine, MSA, RN,
Executive Vice President, Visiting Physicians Association and

David Vezina, MBA,
Chief Information Officer, US Medical Management

According to Joan Valentine, MSA, RN, in an Accountable Care Organization (ACO), “everyone plays nice in the sandbox,” to reduce costs and increase patient satisfaction. This is a very simplified description of an intricate organization — ACOs are comprised of various healthcare providers that work together to provide coordinated care to Medicare patients.

ACOs are required to demonstrate quality through four domains:

  • Patient experience of care: Patients are given the Consumer Assessment of Healthcare Providers and Systems.
  • Care coordination and patient safety: ACOs are assessed based on outcomes measures including ED use, ambulatory sensitive admissions, and readmissions.
  • Preventive health: Standard screenings to increase early detection of various conditions.
  • At-risk populations: ACOs must meet nationally recognized standards for patients with chronic diseases.

Valentine and David Vezina, MBA, worked with US Medical Management as it transitioned to an ACO, which presented unforeseen challenges. Historically, the organization’s patient data had been funneled into various silos, creating barriers for those that needed it. To tackle this challenge, US Medical Management partnered with Health Catalyst to build a single data repository where they could aggregate all of their clinical, claims and financial data. This resulted in $47M in Medicare savings in 2016, making US Medical Management the “fourth best” ACO in the nation.

Vezina ended the presentation by emphasizing the importance of automation, while encouraging attendees to limit the burden placed on physicians as much as possible, saying that their attention should always be primarily focused on providing the best care.

Session 11: Using Analytics to Increase Cash Flow

Greg Stock
President & Chief Executive Officer, Thibodaux Regional Medical Center

Mikki Fazzio, RHIT, CCS
Director, HIM and Clinical Documentation Improvement, Thibodaux Regional Medical Center

Greg Stock opened the session by introducing the audience to Louisiana’s Thibodeaux Regional Medical Center, explaining that the challenge faced there, managing discharged not final billed (DNFB) cases—where bills remain incomplete due to coding or documentation gaps—represents a cash-flow issue for hospitals around the country. Indeed, controlling DNFB is a common strategy for improving financial performance and preserving margin to support hospitals’ care missions. 

Mikki Fazzio described Thibodeaux’s project to understand and address DNFB cases.

  • Identified multiple causes of documentation and coding gaps: these included a slow, manual processes, lack of analytic insight into bottlenecks and issues, and overburdened staff.
  • Implemented interventions: invested in an analytics application that automated a redesigned process and gave visibility into the workflow; additionally, robust education, new departmental structures and roles, and broad access to data all supported the newly streamlined way of working.
  • Measured results: sustained improvement was noted across key measures:
    • 6.2 day reduction in A/R days
    • 51% reduction in delinquency rate
    • 61% reduction in total DNFB
    • $2.4 million in improved cash flow
    • 96% improvement in operational efficiency

The presenters offered lessons and recommendations for others looking to improve DNFB management with analytics:

  • Verify: Take time to validate data before sharing it. Inaccurate data has no value and can breed resistance.
  • Share: Access to clear, accurate data drives change and promotes physician engagement.
  • Take a holistic approach: Data alone won’t solve problems. The formula for sustained improvement is Analytics + Efficient Processes + Accountability.

Breakout Sessions: Wave Two

Session 12: Leveraging Predictive Models to Reduce Readmissions

Rhiannon Harms
Executive Director, Strategic Analytics, UnityPoint Health

Ben Cleveland
Data Scientist, UnityPoint Health

Rhiannon Harms described how UnityPoint Health (UPH) asked UPH Analytics to help the organization reduce readmissions through predictive modeling, with three challenging goals:

  1. Improve predictive model performance from industry standards, such as LACE.
  2. Predict not only readmission risk, but also dates when risk would be highest.
  3. Deliver predictions using an automated, intuitive, cross-continuum tool that everyone on the multidisciplinary team could use.

The team began with a retrospective view, using interviews to identify factors leading to each readmission—especially asking the patient: “why do you think you’re back?” These insights were used to build predictive models, evaluating 114 factors to answer key points of uncertainty:

  • Which patients should we focus on?
  • When is the best time to schedule follow-up for these patients?
  • How should the patient’s compliance with follow-up affect the plan?

Ben Cleveland demonstrated the resulting Readmission Risk Tool, an at-a-glance “heat map” for patients at highest risk. Colored cells show relative risk (green, yellow, red) for each of the 30 days after discharge, so the team can schedule follow-up when risk is highest. The tool also uses a color key to show follow-up appointment compliance, so the team can adjust the plan.

The Readmission Risk Tool is the system’s most widely used predictive analytic tool, model performance compares positively with industry standards, and interdisciplinary teams are using the tool to reduce readmissions. One key lesson: minimize the data literacy required to consume data outputs. Risk scores can trigger worklists, but risk visualizations trigger powerful conversations.

Session 13: Standardizing the Collection of Social and Economic Risk Data

Andrew Hamilton, RN, BSN, MS
Chief Informatics Officer, Deputy Director, AllianceChicago

What determines the quality and length of a person’s life? Clinical care is important, and we have some visibility into this determinant, via healthcare encounter data. But what about factors such as our income level, what we eat, and the quality of our air and water? These social determinants of health (SDH) have been clearly linked to human health, yet they’re outside the traditional boundaries of healthcare delivery.

In this session, Andrew Hamilton described efforts to integrate SDH into clinical care, with the goals of better understanding and serving our patient populations—and better meeting the demands of value-based payment models. He identified steps for integration:

  • Assess SDH needs.
  • Link patients to community services.
  • Use data to evaluate the impact of creating a link between healthcare delivery and community services.
  • Develop sustainable business models to fund access to community services.

Data is critical at each step, but for SDH data to be shared and used, it must be standardized. Mr. Hamilton described the development and pilot implementation of the Protocol for Responding to and Assessing Patients’ Assets, Risks and Experiences (PRAPARE), a national standardized patient-risk assessment protocol designed to engage patients and help health centers and other providers collect and apply the data they need. PRAPARE is:

  • Actionable at the patient and population level.
  • In the EHR (via free templates).
  • A conversation-starter and patient-centered.
  • Consistent, yet flexible: can be implemented in different workflows and made more/less granular as needed.

Six community health center sites have piloted PRAPARE, and early results support the idea that a more-holistic view of our patients—and a more holistic, cross-continuum approach to care—may lead to positive changes at the patient, health center, and community/population level.

Session 14: Using a Real-time Data Science Platform to Drive Perioperative Quality and Efficiency

Bala G. Nair, PhD
Director, Associate Professor, University of Washington

The perioperative space (care in the time period surrounding a surgical procedure) accounts for 50 percent of a hospital’s revenue yet is costly, high-risk, and generates 20 to 30 percent of all hospital waste. As half of perioperative adverse events are preventable, the space presents significant improvement and cost-savings opportunity. Bala G. Nair, PhD, Director and Associate Professor at the University of Washington (UW), explained how UW has used data science to improve perioperative care and efficiency.

Using data and technology to improve quality of care with real-time and predictive decision support and automated algorithms, UW developed and implemented a point of care, real-time decision support system for perioperative care. This solution works with a perioperative information management system to guide clinicians on best practice protocols, deliver high-quality care, and minimize waste.

According to Dr. Nair, the perioperative data platform takes data from EMR in real time, enabling real-time decision support, supply chain optimization, and event prediction. Data feeds automatically into the anesthesia information management system (AIMS) and notifies the clinician of patient needs. In a use case example, real-time perioperative decision support reduced waste of inhalation agents by identifying ways to reduce flow setting at the right times and issue reminders to clinicians to change settings. The addition of real-time perioperative decision support resulted in an annual cost savings of $103,000.

Session 15: Improving Risk Adjustment Coding Accuracy with Analytics

Rod Christensen, MD
Vice President of Medical Operations, Allina Integrated Medical Network, Allina Health, and

Miriah Dahlquist, PT
Senior Performance Improvement Consultant, Allina Integrated Medical Network, Allina Health

In 2015, Allina Health noticed that their patient populations were regularly being reported well below national averages when looking at Risk Adjustment Factor (RAF) scores. Upon discovering this discrepancy, Dr. Christensen and Miriah Dahlquist spearheaded an initiative to improve the accuracy of Allina Health’s RAF data.

The first step in this initiative was to create an interdisciplinary workgroup that included individuals from all facets of the organization. This workgroup developed five key goals for their work:

  • Ensuring an accurate problem list: The most common problems were prioritized while duplicate or extraneous codes were removed; outdated problems were moved to the patient’s history.
  • Ensuring patients are seen in each calendar year: A dashboard was built and included both EMR and claims data which helped to identify patients who hadn’t been seen.
  • Decision support and EMR optimization: Physicians built appropriate coding into their daily encounter workflow when able.
  • Widespread education and communication: Physicians were educated about the coding changes and were encouraged to provide input.
  • Tracking performance and identifying opportunities: The dashboard allows different clinics to compare their RAF data against others in the system, allowing for knowledge sharing and improvement opportunities.

As of today, this initiative has helped to raise Allina’s RAF scores to be on par with national averages. In closing, Dr. Christensen said that he does not yet consider Allina to be a “high-performer” in this area, but says that they are now “average” — with more work to be done.

Session 16: Detecting, Monitoring, and Preventing Patient Safety Events

Robert Quickel, MD, FACS
Vice President, Surgery and Procedural Care, Allina Health

Kassie Ryan, RN, MSN
Improvement Specialist, Health Catalyst at Allina Health

With patient safety being such a widespread problem among healthcare organizations, Allina Health decided to get serious about safety several years ago. While they had good mechanisms for detecting patient safety events, they were falling short because they didn’t have a system-wide approach. They implemented the following specific system-wide strategies make improvements:

  1. Developed a Patient Safety Committee.
  2. Performed a Safety Culture Survey – a key finding was that employees weren’t  comfortable reporting patient safety events.
  3. Hired a COO that had patient safety as a main focus.
  4. Began telegraphing safety events consistently to senior leaders within 24 hours.
  5. Implemented Tiered Safety huddles.
  6. Hired a National Safety Consultant
  7. Examined How they Detect Specific Events

They also partnered with a vendor as a beta tester to develop a trigger tool to be used in specific patient safety events. The trigger tool was based on a specific definition of a harm event that could be customized to the system and pulled from discrete data in the EMR. What they set out to do was use existing reporting to create a trigger tool and ultimately create predictive analytics to identify at-risk patients and prevent safety events from occurring.

Dr. Quikel shared some of the lessons learned on their patient safety journey:

  • Safety systems require accurate complete data to facilitate learning.
  • Voluntary reporting systems provide incomplete safety information.
  • Trigger tool may be useful in retrospectively identifying safety events.
  • Predictive analytics may move us to identify patients at risk in addition to those harmed.

Session 17: Privacy Analytics: A Johns Hopkins Case Study – Using AI to Stop Data Breaches

Robert Lord
President & Co-Founder of Protenus

Robert Lord opened by making the point that data breaches in healthcare are all-too-common. In other industries, data breaches have remained steady over the last decade; not so for healthcare. In fact, data breaches have grown dramatically. Because of this, trust in the healthcare industry is at risk.

When Protenus became engaged in the Johns Hopkins rules-based data breach defense model, their need for more trustworthy systems had become an institutional priority. Johns Hopkins faced several challenges: a feeling they were simply checking the box when it came to HIPAA, lack of comprehensive review, massive amounts of overwork (detecting and reviewing cases), and expanding system access (community connect system, acquisitions, etc.).

Working with Protenus, they determined that a shift in approach to a new platform that enabled detection and resolution was needed. Protenus and Johns Hopkins worked together to:

  • Establish KPIs: threats discovered, false positive rate, burden of tool maintenance, investigation time, and overall number of threats reduced over time.
  • Address organizational challenges: privacy-security silos (integration), fear of monitoring (education), senior leadership buy-in (big picture, ROI, etc.), budget (real and “silent” FTEs, costs, total cost of ownership).

The results of the project were impressive:

  • Minutes spent per case reduced from 75 to 5 minutes.
  • Accuracy increased from 17% to 97%.

The data breach defense model enables Johns Hopkins to accommodate scale (enterprise-wide solution with comprehensive, not sampled, review), complexity (dashboards helped quickly identify outliers), and automation (NLP, automated emails, etc.).

Keynote Session

Jill Hoggard Green, PhD, RN
COO and President, Mission Health

Dr. Jill Hoggard Green started off the Wednesday afternoon keynote session by sharing an incredible story of transformation at Mission Health. The 130-year-old not-for-profit health system was founded by Anna Woodfin in response to the inequality of care she saw in Asheville, North Carolina. Today, the health system faces a number of challenging social determinants of health, including high rates of lung cancer, behavioral health, drug overdoses, and child poverty. Despite these challenges, Mission Health was named one of the nation’s top 15 healthcare systems by Truven Health for the last six out of seven years. It is increasing its numbers served, is more financially sustainable than ever, and deliver increasingly high-quality care

They were able to do this by employing five major strategies:

  1. Become a truly great place to work and practice.
  2. Provide the safest, highest-quality care in the nation when, where and how desired by consumers.
  3. Achieve long-term financial stability.
  4. Achieve targeted growth.
  5. Effectively grow and manage our at-risk population.

Creating a great workplace helped drive many of their other achievements. They found that only 17% of people are engaged in their work. Mission Health implemented a plan with two critical components–leadership and engagement–and saw a 120% improvement. They also focused on providing the right tools that would help drive process improvement with data. Now they’re seeing the transformation in the form of lives saved.

Dr. Hoggard Green shared incredible statistics about their transformation:

  • 37% increase in lung cancer screening
  • 42% reduction in hospital stroke mortality
  • 600 lives saved annually
  • $263 million in total cost reduction

Keynote Session: Achieving the Promise of Digital Health: Are We There Yet? If Not, How Do We Get There?

Robert Wachter, MD
Author, The Digital Doctor, Specializing in Analyzing the Healthcare World in Lively, Iconoclastic, and Humorous Ways

Dr. Wachter initially became interested in health IT when he got his first iPhone. It felt magical. He thought if we had something like that in healthcare, everything would improve.

Instead of eliminating simple problems like the famously illegible doctor’s note, the digitation of healthcare has introduced a whole new set of problems. He gave a shocking example of patient safety gone horribly wrong. Through a series of errors involving healthcare IT, a 16-year-old that was supposed to be prescribed an antibiotic twice a day was instead prescribed 39 pills in one day. He had a grand mal seizure and ended up in the emergency department. Luckily, he survived. Dr. Wachter provided a number of examples and key findings from his book “The Digital Doctor: Hope, Hype, and Harm at the Dawn of Medicine’s Computer Age.”

He pointed out that 10 years ago, only one in 10 U.S. hospitals had an EHR. Today, only one in 10 does not. This illustrates that although the need for rapid technological improvements in healthcare is necessary, we are making progress. While today’s challenge is the shift to value-based care, Dr. Wachter predicted that the next challenge will be digitization of the U.S. healthcare system.

He also shared the Four Stages of Health IT:

  1. Digitize the record.
  2. Connect all the parts (interoperability).
  3. Glean meaningful insights from the data.
  4. Convert these insights into action that improves value.

He posits that we are still in the first stage of this journey. We are not yet seeing the results of this digitization. The two keys to moving beyond this productivity paradox are improvements in technology and a reimagining of our work. Digital is only a component that will allow us to provide better, safer, and more cost-effective care. Though today’s medical students have the huge task to improve the quality and safety of care while lowering cost, Dr. Wachter is optimistic that healthcare if moving in the right direction.

Database vs Data Warehouse: A Comparative Review

This website stores data such as cookies to enable essential site functionality, as well as marketing, personalization, and analytics. By remaining on this website you indicate your consent. For more information please visit our Privacy Policy.