Home Coding Population health 101

Population health 101

June 2015
Female doctor in hospital room, portrait

David Frenz, MD Published in the June 2015 issue of Today’s Hospitalist

POPULATION HEALTH IS SUDDENLY SEXY. It’s one-third of the Institute for Healthcare Improvement’s Triple Aim Initiative. The National Quality Forum is currently working on its own population health framework, and lots of health care systems (including mine) have created accountable care organizations (ACOs). Heck, some hospitals and universities, including here in the Twin Cities, have even created entire population health departments.

But what, exactly, is population health? Where did it come from, and what does it look like in real life?

In the beginning
David Kindig, MD, PhD, and Greg Stoddart, PhD, published an influential article, “What Is Population Health?” in the March 2003 issue of the American Journal of Public Health. They noted that while population health emerged in Canada in the 1990s, there was really no consensus definition at the time. Drs. Kindig and Stoddart then offered their own definition: “the health outcomes of a group of individuals, including the distribution of such outcomes within the group.”

Although this definition seems pretty straightforward, it contains some important nuances. For example, the authors deliberately chose the term “health outcomes” rather than “health status” because “the latter refers to health at a point in time rather than over a period of years.” And the word “distribution” is a nod to health inequality, the fact that some people have better (or worse) health outcomes than others in the same population.

Then there’s the population itself, which could be just about anything. Drs. Kindig and Stoddart indicated that “these populations are often geographic regions, such as nations or communities, but they can also be other groups, such as employees, ethnic groups, disabled persons, or prisoners.”

Finally, this conceptualization includes determinants of health such as “medical care, public health interventions, aspects of the social environment (income, education, employment, social support, culture) and of the physical environment (urban design, clean air and water), genetics, and individual behavior.”

None of this, when you think about it, looks even remotely like hospital medicine. Although our hospitals serve many ill-defined populations, we still treat patients one at a time. We recognize that our patients probably have a range of health outcomes, but we don’t really know the specifics. And social determinants of health appear to be a one-way street: They fill our hospital beds, but we are powerless to change them.

Lost in translation
A lot has changed since 2003. For starters, population health has trickled all the way down from massive government agencies to health care systems. A definition that has been guiding policy wonks isn’t working so well for doctors in the proverbial trenches “or even the ACOs that employ them.

This disconnect prompted Dr. Kindig to revisit the whole subject in an April 2015 blog post in Health Affairs, “What Are We Talking About When We Talk About Population Health?” He concluded that we now need two definitions for population health. He recommended retaining the original 2003 definition “for geographic populations, which are the concern of public health officials, community organizations, and business leaders.” For clinical populations “patients and those treating them “he advocated using terms like “population health management,” or, more appropriately, “population medicine.”

This makes a lot of sense. Federal bureaucrats trying to reduce health disparities are going to approach population health a lot differently than hospitalists. The government can shift dollars, but you can’t deprive Ms. Jones of medications and an MRI because you want to give them to Mr. Smith instead.

Real world
Okay, enough abstraction. 
Let’s pretend that your health care system operates a few hospitals and a bunch of community-based clinics. It recently formed an ACO with a local insurance company and you’ve joined its board as the token hospitalist.

The ACO’s number-crunchers have noticed that a bunch of young, ostensibly healthy members are costing a fortune. Lots of clinic and emergency department visits, hospital admissions, and pharmacy claims. You agree to review some medical records to search for root causes.

It doesn’t take long. The patients are opioid dependent: They’re abusing heroin and prescription pain pills. They’re hitting up primary care doctors and emergency department physicians for opioids. They’re also being hospitalized for overdoses and expensive drug-related misadventures like endocarditis.

You start pulling literature. A study by Howard Birnbaum, PhD, and his colleagues in the April 2011 issue of Pain Medicine found that the annual societal cost for opioid addiction is $56 billion. Health care accounts for roughly half the total, and 95% of that spending is for bad utilization. This is completely consistent with what you’re seeing in your health care system.

You recover some vague memories from medical school about methadone and buprenorphine (Suboxone). A little more digging leads you to a study by Frances Lynch, PhD, and a team at Kaiser Permanente, published in the August 2014 issue of Addiction Science and Clinical Practice. They found that untreated opioid addiction costs about $31,000 per patient per year. That number drops to just $14,000 when patients receive addiction counseling plus buprenorphine. This comports well with many prior studies that demonstrated the same.

But cheaper isn’t always clinically better (dead people have very low utilization). So you continue to root around and eventually land on a review article by Jessica De Maeyer, PhD, and collaborators in the September 2010 issue of the International Journal of Drug Policy. They noted that multiple studies have shown that methadone and buprenorphine quickly improve quality of life for opioid-dependent patients and that these benefits persist over time.

Happy ending
This population medicine fairy tale ends with you urging your ACO board to fund an outpatient infrastructure for addiction counseling and buprenorphine. The number-crunchers love the cost savings. The clinicians are delighted with the improved health outcomes. And, truth be told, you had fun working on the project even though the final solution had little to do with hospital medicine.

Ready or not, this is what your practice (at least to some degree) will eventually become. Every patient that you see for heart failure, hip fracture or pneumonia is a population signal; each of them represents the entire population that you serve. What can you do to prevent these diseases in that broader population or to favorably alter their natural histories once disease has developed?

And with ACOs and big data, things will increasingly be happening the other way around: The population signals are also buried in your EHR and claims data. What can you do to find and interpret them?

David A. Frenz, MD, is a hospitalist for HealthEast Care System in St. Paul, Minn., and is board certified in both family and addiction medicine. You can learn more about him and his work at www.davidfrenz.com or via LinkedIn.