Berkeley Economics Provides Foundation for Unique Alumni Contributions to Health System Optimizations

Alum Marc Tobias (2000, B.A. in Economics) became interested in healthcare delivery during his time at Berkeley Economics, and went on to become one of the first clinical informatics fellows at an accredited program in the country. His research was focused on decision support and inefficiencies in healthcare workflows and processes. He spun this research out as an independent company in 2019 called Phrase Health. In August 2021, the company received a $1.7m NIH grant to help drive improved clinical outcomes through the use of process analytics and quality improvement methodologies. He shares his journey and what's ahead.

Marc Tobias, MD; Phrase Health

If you’d asked me 21 years ago - as I was arriving at Berkeley for the ﬁrst time as an undergrad - where I’d be today, I’m not sure I’d ever have predicted that a casual interest in building websites and a deeper curiosity about system eﬃciencies would’ve led to becoming an emergency physician and running a healthcare quality improvement company. Indeed, my exploration at Berkeley as an Economics undergraduate was instrumental in forging the path I’m on today, leading Phrase Health in our eﬀorts to reduce operational burden, prevent clinician burnout, and improve patient outcomes nationwide.

My path to Berkeley began in an unassuming way. I really wanted to attend a big school. I grew up in Southern California and the UC system has a great reputation at an aﬀordable price. My older brother attended Berkeley (and subsequently, my younger sister did too - all my siblings are Golden Bears!), so I had some personal experience with the institution. I had enjoyed my AP Computer Science classes in high school, so the reputable engineering program was attractive to me. The other programs were well-rounded too and thus, Berkeley was the complete package.

After arriving in 2000, I took the CS61A course (is that still taught in Scheme?) and realized I was more interested in computer science as a hobby than a “science”. I explored Cognitive Science as a possible area of pursuit, driven by my interest in understanding computer science in the context of biology, but settled on Economics after taking an introductory class and recognizing its application to virtually every system we interact with.

I found it fascinating that you could model behaviors to better understand why things are the way they are in markets and everyday life. It was (and still is) a skillset that is applicable to so many domains. I have continually applied the concepts of supply constraints on ﬁxed systems to great avail; it has been relevant to me as a small business owner ﬁguring out the right time to hire a new employee, and yet again as a physician as I theorize strategy on how to juggle a crowded emergency department. In terms of applicable knowledge, economics seemed then and has proven to be, a deep well of applicable knowledge.

I became more interested in biology and medicine towards the end of my college career. At that time, I remember taking a Healthcare Economics course from Dr. Ted Keeler in which we learned about the unique circumstances of healthcare delivery in the United States, but also abroad. This really piqued my interest in the complex and ineﬃcient markets of healthcare delivery.

I ended up pursuing medical school and an emergency medicine residency. After that, I was accepted into a two-year clinical informatics fellowship at Children’s Hospital of Philadelphia. Clinical informaticists are clinicians that are trained to understand, among other things, how to implement processes that eﬃciently and eﬀectively deliver optimal care to patients. Since today’s clinical processes revolve around electronic health record systems, a deep understanding of healthcare data standards and available technological tools are important.

This new board-certiﬁed specialty melded my interests in engineering, medicine, and economics. It was at this point that several of my paths began to converge, creating opportunity for a deeper, more speciﬁed area of impact.

Given my interests, I focused my research on healthcare data interoperability and decision support in the context of quality improvement. More speciﬁcally, I enjoyed applying my interests in software development to electronic health record interoperability, while utilizing my understanding of economics to deliver more eﬃcient solutions to health systems. My early work in the fellowship led to research software called PHRASE (Population Health Risk Assessment Support Engine), which won several awards and began to gain traction operationally within the hospital system.

As I explored this space further, I became very interested in quality improvement. How could we use electronic health record data and tools to inform workﬂows that enable healthcare systems to reach maximal operational eﬃciency and eﬀectiveness? It was a relatively new take on a familiar hope. The shared quality goal in healthcare is the Quadruple Aim:

1) Improving patient experience

2) Improving patient health

3) Improving clinician work life

4) Ultimately reducing cost

I found great promise in a domain called clinical decision support (CDS). CDS encompasses any reference material that helps clinicians make better decisions. This can be as simple as a website that a clinician visits when they have questions about a patient they are treating. More

advanced CDS systems are embedded within the electronic health records and include interventions like alerts and order sets that are personalized to a speciﬁc patient of interest. An alert, an example of CDS, is a visual notice that appears on screen (similar to a pop-up advertisement appearing on a website) which is automated and triggered to appear based on the information in the electronic health record, contextually relevant to that patient. Alerts in the health record system can interrupt the clinician about a potential quality or safety issue (e.g. ,“are you sure you want to order penicillin in this patient with a penicillin allergy?”) or simply provide a gentle reminder about a step that needs to be taken (e.g., “provide at-home care instructions prior to patient discharge.”)

Likewise, another CDS example, an order set is a curated list of orders that can be used within speciﬁc clinical workﬂows in order to reduce cognitive burden, limit the number of clicks required, and decrease care variation. As an example, imagine two physicians in the emergency department. Both may see a patient with stroke symptoms and, depending on their training and experience, will likely kick oﬀ dissimilar orders during triage. However, with an established stroke order set, both clinicians can execute the same bundle of orders and apply their diﬀering knowledge to a more standardized dataset, thus increasing their relative chances of delivering care that aligns with the most up-to-date best practices.

Of course, none of these tools are panaceas and their custom implementations are rife with issues that can result in clinician frustration and poor adoption. During my fellowship, I began research in this space and recognized that many health systems struggled to manage these tools, but also struggled to ensure they were driving their expected value.

We found frequently that informatics and IT teams would build new tools within the clinical workﬂows, but rarely had the bandwidth or expertise to evaluate their eﬀectiveness and ultimate impact on downstream measures. Likewise, quality improvement tools were frequently requesting new or modiﬁed workﬂows, but had diﬃculty assessing the relative impact of these tools once implemented. The degrees of missed opportunity were obvious, and to me, diﬃcult to resist.

After seeing adoption of my research at a few local health systems, my research was spun out as a small company from Children’s Hospital of Philadelphia in 2019. Our core technology helps health systems manage and govern their CDS knowledge library, while also providing insights into its performance. The company, Phrase Health, has continued to grow since its inception and the team has taken on diﬃcult and entrenched problems in healthcare. In

August 2021, we received a two-year Phase 2 NIH grant to study the link between CDS performance and quality improvement. This study, “Multi-Institutional Implementation of Quality Decisions: a Clinical Decision Support Analytics Tool to Drive Outcomes,“ takes a novel approach to the evaluation of healthcare improvement eﬀorts. As the healthcare industry transitions towards value based care, rather than fee for service, this technology is more important than ever.

Although I didn’t pick it for this reason explicitly nor see the potential when I ﬁrst stepped on campus, it’s clear in retrospect that Berkeley was essential in my ability to forge this path at all. My time there provided me the opportunity to explore a wide variety of interests, but the Berkeley Economics department formed a foundation that acted as a springboard to pursue my current pursuits. I have managed to take many of my learnings and interests - so many of which were formed in those years at UC - and link them together in a career path that is driving real change in the healthcare industry. These concepts of eﬃciency and optimization, engineering, and decision making support are all core building blocks of Phrase Health; in combination, they’re aﬀecting real change by improving clinical outcomes and clinician experiences not just at the 30+ hospitals we are supporting today, but at every health system we branch out to next. And what better time than now to support quality, eﬃcient, and less burdensome healthcare?

Marc Tobias became interested in healthcare delivery during his time at Berkeley Economics, and ultimately completed an emergency medicine residency, and then went on to become one of the first clinical informatics fellows at an accredited program in the country. His research was focused on decision support and inefficiencies in healthcare workflows and processes. He spun this research out as an independent company in 2019 called Phrase Health. In August 2021, the company received a $1.7m NIH grant to help drive improved clinical outcomes through the use of process analytics and quality improvement methodologies.