Measuring Population Health Results – What Works?
Health care costs’ going up is old news, yet year after year, it commands the headlines. Some years, “moderate” increases are celebrated – such as 4% in 20151. Let’s put that into perspective. A four-percent increase on family health premiums in 2015 is more dollars ($711) than the 11 percent increase was in the year 2000 ($647)2.
A four percent increase is a heavy burden. Anyone who pays for health care or health insurance – employers, governments, households – has a keen interest in curbing costs. So, lots of things have been tried. See Table 13 for a list of popular strategies. Half of the listed strategies have been shown in research to make a difference and reduce costs. All were undertaken with the goal of lowering costs, and not only did half fail, but some also added costs!
Table 1 – Selected Health Cost Management Tactics and Research Findings
|Tactic||Makes a Difference/Reduces Cost||Mixed Results/Adds Costs|
|Employers making fixed dollar contribution and offering three or more health plans||✓|
|HMOs – competitive market||✓|
|HMOs – non-competitive market||✓|
|Managed mental health/ substance abuse care||✓|
|Flexible spending accounts||✓|
|Lower-income people: cost sharing (deductibles, co-payments, co-insurance) for||✓|
|Higher-income people: high deductible health plans or Consumer-Driven Plans||✓|
|Wellness programs overall||✓|
Why do things that seem logical end up having unexpected or opposite results? This is one of the questions that we will address in this course. You will learn how to see a program in a broader context of people, places, and things; you will also learn how valid and accurate measures of a program and its impact can help you choose programs wisely.
Health care and insurance purchasers are trying a wide range of tactics, in part because we don’t know which ones will work. This begs the question, how will we know what works when it does work? There’s a complex answer to that question.
In this course, we will look at one part of that complex answer –how to measure results (also known as outcomes) of health programs and products. When we can get a handle on results, we can paint a clearer picture of what works and what doesn’t. This course does not address “value” or other unmeasurable impacts; on the basis of unmeasurable concepts, one could choose any intervention and justify it. Our focus in this course is to connect programs to beneficial and measurable results, such as better health status and less need for medical care.
In Module #1, we will define some basic terms: population, intervention, and outcome. Many missteps occur when these concepts are poorly described. For example, if the population is the entire membership of a health plan, then every member must have an equal opportunity to join or be touched by the program; if the population is health plan members who have a particular diagnosis or event, then it is much more narrowly focused. This matters not just for program design, but also for how results are calculated.
Interventions come in many styles and flavors. It could be as simple as increasing the health plan deductible or as complex as creating incentives for people to choose higher quality medical providers. We will show many different types of interventions, after defining them in Module #1.
Likewise, outcomes can be broad ranging. An outcome could be fewer emergency room visits, less back pain, or better compliance with a drug regimen. We will briefly explore outcomes that aren’t directly related to medical care, such as intention to leave the employer or self-perceived health status.
With the foundation laid by Module #1, we move into Module #2 which will address all things data. This includes data design, by which the data is linked as directly as possible to the outcome being measured. Issues with data gathering and data sources are described as well as access and HIPAA considerations. Having credible, valid data is crucial to creating good measures.
Often, access to data is an obstacle. It needn’t be. Population health measures – measures of outcomes from an intervention – typically do not require protected health information. You might need to choose a different measure, but patient-level/ member-level data can in most cases be avoided. Module #2 will discuss various strategies for managing the need for data.
Module #3 covers how data is composed into valid measures. Data by itself does not constitute measures. If the intervention is meant to reduce hospital stays, simply counting the number of hospital stays does not answer the question; measures have to take into account the underlying population and its dynamics of change. Here we will give examples of invalid measures, and the basics of matching data to published standards and composing measures.
You will meet “measure stewards” in Module #3. Measure stewards are organizations that set standards and provide definitions for outcome measures. For example, the American Thoracic Society publishes a measure about use of spirometry tests for people who have Chronic Obstructive Pulmonary Disease.
Three case studies in Module #4 will give you a chance to apply your new skills and knowledge. These are real world examples of populations, programs, goals, data, and measures. Everything from selecting the population, designing the program, setting the goal, developing data sources, and measuring outcomes will be covered.
In the final exam, all of these skills will be applied. You will define a population, describe an intervention and outcome, choose data sources, and compose a valid measure.
Accurately measuring results of population health programs will help us to support high value programs and eliminate weak ones. Putting money and resource where it will do the most good – in this case, reduce medical costs or boost health assets – will ultimately help us all.
Welcome to the course and good luck!