HIMSS 2015 Day 1: The State of Healthcare Analytics

Day 1 Greetings From the Land of HIMSS 2015 and Healthcare Analytics

Today Dr. John Haughom and I attended multiple HIMSS education sessions that stood out because of their focus on using data and analytics to improve clinical care, patient experience, costs and population health. Here is our summary of the highlights of what we heard.

Managing Population Health with Science, Analytics, and Quality Improvement

By Charles G Macias MD, MPH, Chief Clinical Systems Integration Officer, Texas Children’s

After the opening HIMSS keynote, the day kicked off with a standing-room only presentation by Dr. Charles Macias of Texas Children’s. The session was so crowded that we noticed that HIMSS declared it “FULL”. Afterwards we asked and were told that it was filled to capacity at 1000 attendees. The presentation definitely met those high expectations.

Dr. Macias started the presentation with a personal story of a patient, a 14-month girl with no history of asthma, who had a cough for one day consistent with asthma symptoms, crying loudly with respiratory distress. He asked the audience, “What do we really know about her problem to inform the actions? Is it a short-term episode or a long-term problem, and how should we treat it?” Dr. Macias then purposely left the question open and waited until the end to answer it.

respiratory challenge

Dr. Macias then posed another question about the treatment of hypertension on a steel mill job site and identifying the key factors in determining the clinical decision to treat some but not other patients. He shared 3 reasons: the level of diastolic blood pressure, the patient’s age, the amount of target-organ damage, but asked the audience to name the fourth. The audience suggested several answers, including gender, which Dr. Macias said was usually the most common answer. The actual answer was the year the physician graduated from medical school. His point was that we often think that variation is patient-related, but often times, it can be provider or clinician related. His premise was that minimizing variation in practice can improve the quality of healthcare delivery. Despite healthcare’s scientific foundation, variations in practice can often occur in

  • beliefs
  • interpretations of evidence
  • in response when evidence is lacking

Dr. Macias then showed research indicating that variation in practice is in resulting significant gaps in optimal care (less than 50 percent) being delivered to children in outpatient settings.

indicated care reporting in outpatient

variability in pediatrics

He then showed how Texas Children’s Hospital was taking a population health approach to improving these outcomes. He described the Texas Children’s Hospital approach as using a three-pronged foundation: analytical systems, operational improvement teams, and clinical science.

creating a foundation for improving outcomes

He then gave three examples of how Texas Children’s was using this approach to improve outcomes: appendicitis, diabetes, and asthma.


  • Increased postoperative simple order set adoption rates by 36%
  • Increased postoperative complex order set adoption rates by 9%
  • Decreased length-of-stay by 30%
  • Decreased variable cost by 20%


  • 95% order set utilization
  • Doubled (from 53% to 100%) % of patients receiving IV insulin within 1 hour of receiving insulin order
  • Decreased length-of-stay 36% (3.9 days to 2.5 days)
  • Increased from 15 % to 80% of patient care experiences provided by the CDNs


  • Increased early administration of steroids (within 1 hour) from 55% to 92%
  • 35% reduction in length-of-stay with no change in 7-day or 30-day readmission rates
  • Reduced unnecessary chest x-rays by 30%

Next Steps: Predictive Analytics

Dr. Macias then shared how Texas Children’s Hospital is moving to the next steps of using data for predictive analytics in diabetes, appendicitis, epilepsy and other conditions aligned with clinical care goals and payment reform.

He closed the session by finishing the case study question he first posed. The 14-month girl is his daughter. And while they still don’t know the answer, he illustrated in a powerful way why data and analytics matter not only at a systematic level, but more importantly, at a personal level for the families of all the patients.

Using Perioperative Analytics to Reach the Triple Aim

By Shailesh D. Patel, MD, Medical Director and Lori Beucler, RN, CNO, Susquehanna Health, Williamsport, PA

Susquehanna Health is a multi-hospital health system located in central Pennsylvania with a longtime focus on quality and safety (highest patient safety rating among healthcare providers in Pennsylvania and the fourth highest ranking in the United States).

Because of the importance of surgical services, Susquehanna Health sought to leverage data analytics to identify improvement opportunities in perioperative care consistent with the IHI Triple Aim – improving the care experience, improving population health and reducing care costs.

Their data-driven improvements focused on the following areas:

  • Improving patience experience by creating a more patient-centered perioperative environment, reducing waiting times, improving surgery start times and improving post anesthesia care unit (PACU) throughput.
  • Improving clinical care and outcomes by reducing length of stay (LOS), improving compliance with SCIP measures and reducing surgical readmissions.
  • Improving financial results by reducing costs and eliminating waste.

By leveraging perioperative data analytics and process improvement, Susquehanna Health achieved the following results:

  • A new patient flow model eliminated 90 minutes of wait time.
  • Increased patient satisfaction from 94% to 98%.
  • Standardized key perioperative care protocols.
  • Achieved 100 percentile for all SCIP measures resulting in a Quality Insight Innovation Award.
  • Significantly reduced average LOS – for example, post hip and knee surgery dropped from an average of 4.9 to 3.1 days.
  • Balanced the use of in- and outpatient surgery units to realize greater revenue and higher physician satisfaction.
  • Reduced supply costs – for example, orthopedic implant costs were reduced by $300,000.

Data-Enabled Strategic Resource Allocation in ACOs

By Rishi Sikka, MD, SVP Clinical Transformation and Tina Esposito, VP, Advocate Health Care

As the industry continues to shift to a focus on accountability across the continuum of care, it becomes imperative for organizations to reallocate human and capital resources, ensuring efficient and effective care for an entire patient population. The ability to bend the cost curve will occur only if the right resources are deployed to areas where they are most appropriate.

Advocate Health Care, a 13-hospital system in Illinois with 1300 employed physicians and 4300 aligned physicians, operates the largest accountable care organization (ACO) in the country. The Advocate ACO is managing 528,000 with total healthcare expenditures of $4.4 B. Advocate has experienced a reimbursement shift from 82% fee-for-service in 2010 to 55% fee-for-service population management, 12% global capitation, and 3.2% fee-for-service in 2014.

In order to successfully manage populations and achieve value-producing outcomes, Advocate recognized the need to strategically allocate resources across the continuum of care. This required that they better understand risk within a population, consider new care models, optimally match resources to needs, and enable optimal health. Their improvement initiative focused primarily on two things – 1) managing care transitions well by placing each patient in the most appropriate care venue and 2) optimizing outpatient care management (OPCM).

With any given patient, it should be possible to determine the best care location for them (e.g., hospital, rehab, skilled nursing facility, behavioral health, assisted living or home care). They developed a post-acute care network (PAN) and sought to match patients with similar clinical profiles to the best care location in the network (resulting in the lowest readmission rate) and quantify the readmission risk based on the clinical situation and the placement location.

They determined that 71% of the time their approach and model matched patients with the appropriate care location in the PAN, but 29% of the time there was a mismatch representing a $200 M potential savings in the cost of care. The results demonstrated that they were under utilizing home care and significantly over utilizing acute care venues. They also learned that their greatest opportunity was in better managing the costs associated with the 30-day post-discharge period that appeared to be driven by the availability and utilization of skilled nursing care. By optimizing care for patients managed in SNFs by the PAN, Advocate was able to reduce SNF costs by an average of $2000 per episode of care and decrease SNF LOS by 5 days as compared to patients outside of the PAN.

Advocate also set up an OPCM that was designed to be short term (less than 120 days), focused on potentially preventable events, evidence-based and measurable. Potentially preventable events were defined as those identified by a clinician as most appropriate for care management, where interventions can reduce hospital encounters within a 120-day period, and that could be impacted in a measurable way (e.g., a potentially high cost situation that could be managed effectively to reduce costs). The OPCM model is illustrated in this diagram.


About a dozen clinical conditions were identified that could be impacted in a measurable manner. An asthma example is shown below.

OPCM example asthma

By appropriately allocating resources and optimizing OPCM, Advocate hopes to continue to improve it population health capabilities in the future.

Leveraging Clinical Data for Risk Adjusting Bundled Payments

By Suma Thomas, MD, Vice Chairman Operations and Strategy, Cleveland Clinic and Soyal Momin, MS, MBA, Director of Research Environmental Informatics, Healthcore/Athem, Inc.

Bundled services is a single payment for all services related to a treatment or condition and provides financial incentives to reduce inappropriate care. The Cleveland Clinic and Healthcore / Anthem embarked on a study to demonstrate an innovative way to examine the need for risk adjustment of bundled payments using both payer claims and provider clinical data. They also sought integrate claims data and clinical data to develop a Longitudinal Patient Record (LPR) and apply the LPR to risk adjusting bundled payments.

Traditional risk adjustment approaches tend to rely on healthcare claims data. They do not include the clinical data necessary to allow more accurate risk adjustment. Claims data alone provides information on co-morbid conditions, prescription drugs, health plan costs, and indications of adherence to recommended treatment plan. By adding clinical data to claims data, it should be possible to achieve a more comprehensive, multi-dimensional view of a care episode and better assess risks. For example, claims data cannot assess compliance. Why is a provider or patient not following a recommended treatment plan for a condition? Clinical data can add information regarding risk profile including behavior and family history, interventions tried, formulary details, over the counter medication usage, and additional information that was used to inform treatment decisions.

In this study, a cohort of 426 cardiac (PCI) patients was carefully identified to match a very specifically defined bundle using ICD-9, CPT, SNOMED, LOIC and other coding schemes. An LPR was subsequently built using the integrated clinical and claims data according to the development process illustrated below.

LPR Development Process

The population characteristics were then carefully studied and the ability of the model to assess costs was analyzed. As illustrated in the graph below, the model was quite effective at predicting costs except at very high cost levels.

predicting costs

While this study is not yet complete, a number of conclusions have been drawn and lessons learned.

  • Stakeholder alignment is key
  • Many health systems are not ready from a technological perspective (especially in having an integrated source of data from multiple disparate sources, e.g., clinical, financial, claims, etc.)
  • Trust between parties and in the data must be established
  • Understanding of data complexity and ability to integrate data is essential
  • Development and implementation costs can be high
  • The results can be transformational for the healthcare system

End of Day One: Onward to Day Two at HIMSS 2015

And that wraps up our Day One at HIMSS 2015. We’re excited to see what we learn at Day Two and look forward to sharing it with you tomorrow.

Loading next article...