What Healthcare Executives Can Learn from Military Decision Making

Editor’s Note: Dale Sanders, Health Catalyst’s Senior Vice President of Strategy, presented to other executives at the invitation-only 2014 Plante Moran Healthcare Summit. Below is an executive summary of his session, a version of which originally published at Plante Moran Healthcare Executive Summit Highlights.

What can healthcare executives learn from military decision-making, as it relates to using healthcare data? As it turns out, quite a lot.

Dale Sanders, senior vice president for strategy at Salt Lake City, Utah-based Health Catalyst, drew some surprising parallels between these two vital sectors of the economy during the Plante Moran Healthcare Executive Summit on June 5 in Chicago.

Before entering the field of healthcare analytics in 1997, Sanders assessed nuclear threats, first for the U.S. Air Force and later for the National Security Agency and Joint Chiefs of Staff.

“Doctors go through the same decision-making processes as military generals do,” Sanders said. Both leaders must diagnose a situation, seek out relevant data, and determine the intervention that will best mitigate the risk, given time and resource constraints.

Some of the questions that leaders in both settings must ask themselves include:

  • Which targets (organ systems) are under attack?
  • What are my sensors (lab results) telling me?
  • What are my human intelligence sources (patient, family members, nurses) telling me?
  • What forces do I have at my disposal?
  • What do I expect as an outcome?
  • What is the appropriate response?
  • How much time do I have to decide?

Fast, predictable turnaround—whether of aircraft, patients, or supplies—is vital in both environments. Just as a military operation depends on a precisely timed order of battle, a hospital is responsible for delivering the quantities and types of resources necessary to return a patient to a healthy life as fast as possible.

Predicting patient risk is another area where healthcare leaders can learn from the military. Military risk is a function of the probability of attack multiplied by the probability of success and the consequences of that attack. The magnitude of that risk is then measured against the cost to intervene.

“I don’t see a lot of evidence of this approach to risk mitigation yet in health care,” Sanders said.

Healthcare providers and policy makers must consider what it will cost to intervene before undertaking expensive analytics. Sanders was recruited by Northwestern University’s medical school to serve as the CIO and chief architect of their enterprise data warehouse, which provided data to researchers, clinical quality initiatives, and internal management reports. “We found at Northwestern that, unfortunately, some of our highest-risk patients in some of our disease registries were unlikely to change their risk status, no matter how much clinical time and management we invested in them,” Sanders said. “You can predict all you want, but it’s the intervention that matters. And if you don’t have a model for cost-effective intervention and mitigation in your strategy for risk management, you are going to be up the creek without a paddle.”

Speed of Healthcare Software

“Like it or not, fast or slow, your company now adapts to change at the speed of software,” Sanders said.

When he joined Intermountain Healthcare more than 15 years ago as regional director of medical informatics and chief architect for the enterprise data warehouse, Sanders began developing a roadmap for healthcare data analytics, which is now known as the Healthcare Analytics Adoption Model. This model was formally published in 2012, with the help of co-authors Denis Protti and Dr. David Burton. It purposely borrows from the success of the HIMSS Analytics EMR Adoption Model.

healthcare analytic adoption modelSanders pointed out that organizations must walk before they can run, and healthcare organizations should get some basics in place before attempting predictive analytics.

“There is probably not a topic that is more over-hyped in healthcare IT right now than predictive analytics,” Sanders said. In fact, predictive analytics doesn’t even show up until Level 7, Sanders pointed out. “There is a lot you can and should be doing with analytics, first,” he said. “Some of the most valuable predictions don’t need a computer algorithm. We already know what we should be doing to avoid readmissions. We don’t need a computer to tell us that.”

In addition to readmissions, wasteful healthcare spending is another area of low-hanging fruit, Sanders said. “Waste elimination comes back to your bottom line, no matter what your reimbursement model is,” he pointed out.

Bringing the Patient Picture Into Focus

Whether the purpose is to drive out waste from the healthcare delivery process or improve clinical outcomes, healthcare organizations need greater volumes of data and more diverse types of data.

Consider this: Healthcare organizations collect about 100 MB of data per patient per year, and a Boeing 787 collects 500 GB of data during one six-hour flight, Sanders said. Trying to make accurate predictions with such sparse data is close to impossible. “Like a poor resolution jpeg, the image of the patient will not be clear until we fill in this data picture,” he said.

Currently, most of the data that healthcare providers collect relates to the healthcare encounter itself. But visits to the physician or hospital represent only a small slice of the patient’s life. Healthcare providers should look beyond those encounters to other information about the patient, such as purchasing habits, genomic data, and level of activity.

Biometric data regarding everything from activities of daily living to blood pressure is becoming increasingly more available through wearable technology and software such as the upcoming Apple Health, which promises to integrate data from various health apps and make it available to healthcare providers.

Predictive Precision vs. Data Content

predictive precision vs Data Content

Establish a Healthcare Data Acquisition Plan     

Where Can We Start With Predictive Analytics?

Big data can be overwhelming, but healthcare providers can use a number of factors to decide where to start investing time and money when it comes to predictive analytics.

  1. Can you influence the outcome? Predictions without interventions are useless, Sanders said. Focus your organization’s time and money where you have the greatest potential for influence. “The big gains in clinical outcomes and cost tend to be in more inpatient-centric environments,” Sanders said. For example, common predictive analytics applications being marketed today focus on issues such as readmissions, ulcer management, inpatient mortality, intensive care admission and C-section births.
  2. Is the data available? In addition to control over the patient’s environment, hospitals and other inpatient providers have greater control over the collection of data.
  3. Is intervention cost-effective? Just because you can predict something doesn’t mean you should. “Before you undertake predictive analytics, be sure you have a clear understanding of what you are going to do to intervene,” Sanders said. “If you don’t have a realistic opportunity to intervene, or if you can tell it’s going to cost too much to intervene, stop. Why waste your time on predictive technology that is not going to lead you somewhere productive?”
  4. Do we have the right leadership? Healthcare systems can make tremendous improvements in quality and cost with relatively little data, as long as they have cultural alignment around the need to use that data as the basis for interventions, Sanders said. “At Northwestern, we went from nothing to Level 6 in the Healthcare Analytics Adoption Model in three years. You can do it with the right kind of leadership and commitment.”

“It’s amazing to me as I travel around the country, we’re still thinking about acquiring people and facilities and patient volume, and we’re not thinking about acquiring data,” Sanders said.

He urged the healthcare executives at the Chicago summit to appoint a data governance body, whether a subcommittee of the board or executive committee that is charged with creating a strategic plan for acquiring these data sets.

He also challenged them to set an aggressive timeline—such as five years. “Look at the pace of change that is being demanded of the industry,” he said. “I don’t know how much longer we can continue these slow failure cycles. Employers, our government and the economy are demanding something faster.”

A number of hurdles keep the healthcare industry from fulfilling its population health potential, Sanders said. One of those hurdles is a limited scope of influence.  He referenced an analysis by the Robert Wood Johnson Foundation that found clinical care influences about 20 percent of the patient’s health outcomes. Social and economic factors, from safety to education, have the greatest influence.

“If you are going to be an accountable care organization, where do you draw the boundaries of responsibility for accountability? Does it extend to the patient’s home and lifestyle?” Sanders said. “Where do we grow as healthcare providers?”

Healthcare IT is particularly underused in the long-term care setting, where it is primarily used for state and federal payment and certification requirements. “Until the long-term care environment becomes more computerized, it is going to be hard to risk stratify and target the patients who represent the highest opportunities for improvement,” Sanders said.

Providers and payers also must take into account that certain socioeconomic factors might preclude patients from actively participating in their own health care. In addition to developing different interventional strategies for patients in those categories, the physician compensation models should reflect those limitations.

Speeding Into the Curve

The biggest impediment to investing in healthcare analytics is that the vast majority of healthcare dollars still are spent to reward volume.

Sanders encouraged healthcare executives to embrace the fee-for-value revolution because it will produce better care for lower costs. He shared the example of the Cayman Islands National Health System, where Sanders was CIO for three years. The CEO of the health system proactively approached the nation’s government and commercial payers with a plan to move to a per capita reimbursement model, which was wildly successful.

“I don’t know of a healthcare CEO in the United States who would have the courage to go to the payers and say, ‘Let’s capitate reimbursements,’” Sanders said. “I sure wish leadership in this country would try to push us down this curve faster.”


PowerPoint Slides

Would you like to use or share these concepts?  Download this presentation highlighting the key main points.

Click Here to Download the Slides

Loading next article...