From Elephant to Beehive: Modeling team-based primary care in Utah

By Anne Fischer, Senior Director of Emerging Analytics with IBM Watson Health in Ann Arbor, Michigan, and an IBM Health Corps team member from the Utah Medical Education Council project.

I recently had the extraordinary opportunity to participate in the IBM Health Corps on a project designed to ensure high quality delivery of primary care within the healthcare ecosystem of the state of Utah. The experience was challenging yet incredibly rewarding. I’m excited to share some of the details of the work we performed and the future direction the project may take!

In the struggle to contain the rising costs of healthcare, primary care is uniquely positioned to contribute by helping reduce the incidence and/or progression of chronic disease, reducing unnecessary expensive specialist and inpatient care, and building relationships with individuals to help them focus on a healthier lifestyle. However, Primary care in the United States is facing significant challenges. While the importance of primary care in reducing overall cost is generally understood and accepted, the current fee-for-service based reimbursement mechanisms do not reflect this value. In addition, primary care physicians bear an ever-increasing burden of electronic documentation, coordination among other providers, and general population health management. It’s not surprising that primary care is becoming a less desirable physician specialty. In fact, according to the Association of American Medical Colleges, by 2030, the US will experience a shortage of potentially more than 43,000 primary care physicians.

Utah is no exception to this challenge. The Association of American Medical Colleges ranks Utah 49th out of the 50 states in the supply of primary care physicians per 100,000 residents. While Utah has a younger, healthier population than the average U.S., there are very distinct sub-populations with high primary care needs, particularly in rural and lower-income areas of the state. Mental health is also a significant concern in Utah; the rates of mental health conditions are above national averages and suicide is currently the #1 cause of death among Utah adolescents and 5th across all age groups (Annor FB, Zwald ML, Wilkinson A, et al. Characteristics of and Precipitating Circumstances Surrounding Suicide Among Persons Aged 10–17 Years — Utah, 2011–2015. MMWR Morb Mortal Wkly Rep 2018;67:329–332. DOI:

In 2015, two groups in Utah came together to address the growing primary care challenge. The Utah Medical Education Council (UMEC) and the Utah chapter of the Area Health Education Center (AHEC) began gathering data and creating a strategy that would both inform stakeholders of the current and anticipated gaps in primary care, and also provide direction on how those gaps might best be filled. The goal of the project was to bring together two related but separate initiatives. The first was to measure the anticipated primary care needs of the population of Utah, and to understand where there were provider shortages (and how these might change over time). The second was to promote the use of primary care delivery teams rather than focusing only on individual primary care physicians. These two concepts overlap in that the anticipated gaps in primary care supply might be more efficiently addressed through team-based care rather than simply defining the gaps in terms of physicians required.

The IBM Health Corps became involved in 2018 when UMEC/AHEC applied for a grant from IBM to further their research and serve as a catalyst to bring the project to the next level. The grant was awarded, and six IBMers from around the world found themselves flying to Salt Lake City on the weekend of October 14/15 to engage in a three-week intensive project. I was among those six individuals.

UMEC and AHEC had already done a considerable amount of work to lay the foundation for solving this problem. They had gathered data pertaining to the current primary care workforce and its geographic distribution, the Utah population and its anticipated growth trends, and the types of primary care needs for a given cohort of patients based on recommended guidelines in three areas: acute care, chronic care, and well care. One of the most challenging data assets that they created required a component analysis of various primary care visit in which specific tasks were itemized, estimated, and assigned to certain provider types. For example, most office visits consist of patient biometric collection (height, weight, blood pressure, etc.). The team had itemized the components that are a part of this biometric collection process, how long those component tasks typically take, and what type of provider role was able to perform those tasks given their training and licensure. When the Health Corps joined the project, we were lucky to start with these valuable data assets. Our goal was to create a way to allow these data to answer more complex questions, as well as to make the inputs and resulting answers more transparent, flexible, and accessible to a broader audience.

Our first week was spent “drinking from the firehose” as we learned incredible amounts of information from various stakeholders who were vested in the outcome of our work. We met with representatives from various professions (Physicians, Physician Assistants, Nurse Practitioners, Medical Assistants, etc.), legislators, educators, and public health resources. Using IBM Design Thinking approaches, we were able to translate the various inputs into four clear user archetypes: Legislative, Clinical Administrative, Education, and Health Researchers. These four groups of people all needed a trusted, centralized source of information that would clearly identify the primary care gaps and the composition of provider teams that could fill those gaps in the most cost-effective way. Our shared vision was stated as this: “Ensure that the people responsible for primary care delivery within the health system—at an education, policy, research, and healthcare delivery level—can understand population needs, where there is a gap, and feel confident about making informed decisions on effective next steps to build a primary care workforce.”

Through our many interviews and input sessions, we also learned other important caveats which added both value and complexity to the solution. First, we learned that many of the stakeholders were concerned that providing the “lowest cost solution” would also mean a drop in the quality of primary care being delivered, and in both provider and patient satisfaction. Since the quadruple aim in healthcare is to decrease costs while improving outcomes and patient/provider satisfaction, simply optimizing a staffing solution on cost would not be sufficient and could even be detrimental. Secondly, we learned that team-based primary care would not always be a possibility, particularly in rural areas of the state where primary care physicians often play a much larger role in delivering care, and other types of providers are not often available.

Taking these two additional insights into consideration, we went to work on a solution that would allow for flexibility in defining what a “team” could and should look like as well as what tasks were most suitable for various provider types in a given situation. The goal was to minimize cost, but also maximize a somewhat intangible feature we referred to as “suitability”. In other words, put the “best person for the job” in place as often as possible, while minimizing the cost involved in delivering the care, understanding that the “best person for the job” could vary widely depending specific circumstances in a given area.

Our goal was not to build a complete end-to-end product in the three weeks we had. Instead, we wanted to provide a sophisticated framework that would allow UMEC and AHEC to continue to expand upon the work we had done in the future. We focused on four main areas:

  • create an expandable data framework that simplified the inputs required and eliminated redundancy
  • create a statistical model that took the various inputs, calculated primary care needs, and used parameterized assumptions to calculate the optimized workforce to meet those needs, balancing between cost and suitability
  • define the ideal end-user experience that would facilitate interaction with the model, allowing for “what if” analyses in which a user could change the underlying assumptions and data to understand the impact resulting from those changes; begin building a UI to support that experience
  • Create a technical infrastructure that linked the data, the optimization model, and the eventual UI

Health Corps projects require that all tools and “software” created cannot utilize any IBM proprietary assets. The solution must be able to be expanded upon and maintained by the Grantee after the end of the engagement in a free, open environment without requiring licenses or subscription fees. As such, we used all open-source tools and technologies to build our solution.

So, did we succeed?  How does one measure success in this type of project?  Both the IBMers and the stakeholders learned a lot, and had some fun as well. There was some irony in the fact that we were an interdisciplinary team creating a solution that promoted interdisciplinary teams in primary care. I personally was reminded of what it’s like to work intensely on a single project for an extended period of time and the gratifying experience that can be. Professional relationships and even friendships were formed that I hope will continue. One key stakeholder has repeatedly noted that “we got further with IBM in three weeks than we would have in three years on our own”. That, to me, is success. Did we complete an end-to-end working prototype by the end of the last day?  No. However, while that might have been a stretch goal, the more important goal that we did meet was leaving behind a well-documented flexible framework for the UMEC and AHEC team to build upon.

We presented our work on Thursday of the final week in Utah at the State Capital building. Perhaps it was my lack of sleep, or the high level of adrenaline on that last day, but I felt that the level of enthusiasm in the room for moving the work forward was palpable. Senator Ann Millner of Utah, who had been involved in some of the UMEC/AHEC work prior to our engagement, was in attendance. Afterwards, she approached our UMEC representative and asked him what kind of resources he needed to continue this work because she believed it to be so critical. If UMEC and AHEC can build on the work we started, the population of Utah has a chance to buck the trend and become an example of primary care delivery for the rest of the country. I’m grateful to have had a chance to contribute to such an important effort and I look forward to seeing its long-term impact.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s