This week, Clinical Informatics News is exploring those intangible, sometimes invisible, but altogether indispensable pieces of the health care system – the patients themselves. How can data-driven methods be brought to bear on patients’ lives outside the clinic? And if we ask the right questions, can they tell us how to provide better care?

By Aaron Krol

October 15, 2013 | One of the hottest topics in health care reform today is “patient engagement” – the idea that people who take an active role in understanding and planning their care management can improve outcomes while lessening the work burden on their physicians. It’s a simple, intuitive notion, and like many simple, intuitive notions, has sometimes been resigned to the cabinet marked “Important – Deal With When Measurable.” And a number of obstacles do suggest themselves to collecting useful data in this arena. What factors are involved in engagement, and how should they be weighted? Is understanding one’s drug regimen, for instance, more or less important than asking questions of one’s primary care physician? And can all this be measured accurately enough to grade patients on a meaningful scale?

But for a certain breed of researcher, it’s long past time to start answering these questions.

A New Measurement Tool

Dr. Judith Hibbard, Professor Emerita at the University of Oregon’s Department of Planning, Public Policy and Management, is one such researcher. In 2004, she was co-author of a study published in Health Services Research that proposed a Patient Activation Measure, or PAM, to assign users a simple numerical score based on their responses to a short, highly generalized series of survey questions. Dr. Hibbard knew that, if this measurement tool could be made sensitive and predictive enough, it would give health care organizations the validation they needed to fold patient engagement into their practices. Speaking at the national healthy policy institute NEHI's Innovation Conference this October, she observed that, in medicine, “we measure what matters – and what’s measured is what gets attention.”

Dr. Hibbard (L) speaks about PAM at the NEHI Innovation Conference on Oct 3. Image credit: NEHI

The 2004 paper brought a serious level of rigor to measuring activation, initially defined as having the beliefs, knowledge and skills necessary to manage a chronic illness. The idea was to take the profile of a "highly-activated" patient – someone who believes their role in health management is important, seeks out the best information, communicates constructively with care providers and follows through with their physician's recommendations – and devise test questions that would accurately place individuals on a sliding scale relative to that ideal profile. To that end, the researchers solicited dozens of test questions from an expert panel, administered those questions to patients with chronic conditions, and used statistical tools to winnow the test items down to the most predictive.

One of the first promising signs that PAM was digging at a real quality was the observation that the survey items could be organized on a Guttman-like scale. This means that positive responses to more “difficult” items – i.e., “I have made the changes in my lifestyle like diet and exercise that are recommended for my health condition” – strongly implied positive responses to less difficult items like “Taking an active role in my own health care is the most important factor in determining my health and ability to function.” The reverse, however, was not true: responses to less difficult items did not predict the responses to more difficult items. This Guttman-like organization demonstrates that people can be sorted into a meaningful ladder, with activation occurring in steps that must be achieved in order. In PAM, there are four steps on the ladder, which the authors defined as “believing the patient role is important;” “having the confidence and knowledge necessary to take action;” “actually taking action to improve and maintain one’s health;” and “staying the course even under stress.” A respondent’s PAM score would reliably assign them to one of these four levels of activation. “We think it is actually a measure of their self-concept as a manager of their health,” Dr. Hibbard told Clinical Informatics News.

PAM was identifying a true characteristic of the patients tested, but this would have little value if it didn’t translate to more traditional measures of health. The next step was to see whether PAM measures tracked with health outcomes over time, and for this, the authors of the study would need partners with a wider reach. Armed with a few initial correlations – showing, for instance, that high PAM scores were associated with better scores on the widely-used SF-8 survey of self-assessed health – Dr. Hibbard and her colleagues were able to convince organizations like the AARP, Fairview Health Services in Minneapolis, and the Kaiser Permanente Center for Health Research to adopt the measurement tool for large-scale patient studies. Since then, a staggering array of health variables has been shown to correlate with PAM scores: exercise, diet, smoking, preventive screens and immunizations, substance abuse, obesity, emergency room visits, levels of triglycerides and LDL, and medication adherence, to name just a small sampling. PAM scores even predict better outcomes from acute illnesses, despite the test originally being devised to measure how well an individual could manage a chronic condition.

Even more significant is the realization that a patient’s PAM score can be used as a plan of action. Long-term studies of cohorts who take the PAM have demonstrated that scores are not static, but can be moved up or down depending on outreach efforts and life events. This opens up the possibility of targeting education and self-management tools based on each individual’s level of activation. “Some organizations are actually thinking about their whole population, and reaching out in a more proactive way to patients who are less activated and have higher disease burdens,” says Dr. Hibbard. Because low PAM scores are easiest to improve, and because improving health behaviors from a lower baseline produces more dramatic results, this can be a very efficient way to target health investments.

Dr. Hibbard offers the example of a hospital following up on an admission: “When patients are discharged, [hospitals] will send someone to the home if the patient is low-activated… They go into the home, make sure they have their medications, make sure they’ve got appointments with their doctor.” With a higher-activated individual, meanwhile, the hospital might get the same result from a simple phone call, saving time and funds without compromising health outcomes. To supplement hospitals’ outreach efforts, Insignia Health, the Oregon-based company that licenses PAM, has also introduced an online tool called Flourish that offers health coaching and educational materials tailored to an individual’s PAM score and the chronic conditions they’re managing. The patients, meanwhile, often prefer this kind of evaluation to the typical forms they fill out in the clinic’s waiting room. “People have been very positive about the items themselves,” says Dr. Hibbard, “because they’re different. They’re not like, ‘Well, have you been smoking or drinking?’ They’re not very threatening.”

Rapid Adoption

PAM has now been translated into 20 languages, and new correlations with PAM scores continue to be documented at a rapid pace. Just this October, Dr. Suzanne E. Mitchell and a team at the Boston University School of Medicine published a study showing that the lowest-activated patients were 75% more likely than the highest-activated patients to be readmitted to a hospital within 30 days of a discharge. Dr. Hibbard herself cooperated with Fairview on a study published this February, which worked with a patient population of over 30,000 to show that the lowest-activated patients could represent an increase in health care costs of 8-21% over their highly-activated counterparts.

With so much potential to impact the cost and outcomes of care, PAM is drawing increasing interest from the insurance industry. Companies like Medica, United Healthcare and Wellpoint are providing incentives to their customers to take the PAM or complete challenges associated with improving their scores. With the rollout of the Affordable Care Act, insurance companies are even using their new populations’ PAM scores in the absence of claims data to help determine what rates they should charge.

Of course, the obstacles to incorporating patient engagement into clinical care won’t evaporate overnight. Working on PAM scores can give individuals the perspective they need to make considered choices about their care, but engagement remains a complex relationship between patient and provider. “Both physicians and nurses are trained basically to tell people what to do,” says Dr. Hibbard. “And yet we know that telling people what to do doesn’t work. So the idea that it’s part of your scope of work to support patients in gaining confidence, and skills, and knowledge – that is a paradigm shift.” To fully capitalize on patient activation, clinics and hospitals will need to bring new skills to the table, whether by hiring outreach professionals, or by incorporating patient engagement into the medical education system. Still, the benefits of such a sea change would flow both ways. “I actually think there would be less clinician burnout and unhappiness if they had these skills,” Dr. Hibbard adds. “A big source of unhappiness for clinicians is, you know, ‘My patients don’t do what I tell them.’ And they feel ineffectual, and disempowered.”

A reimagining of that relationship could cast PAM in a new light, as physicians seek not only to give their patients a laundry list of medications and behaviors, but also to activate them enough to ensure adherence. In this vision, PAM isn’t only a predictor of outcomes – “it could be an outcome of care,” says Dr. Hibbard, “an intermediate outcome of care that can be measured and tracked.” In the long term, physicians could even be evaluated on their ability to increase activation. And if it seems like a burden to make physicians responsible for their patients’ willingness to pursue information and healthy behaviors, just remember that even the earliest stages of activation are all about assuming control over one’s own health.

In fact, it’s the very first item on the test: “When all is said and done, I am the person who is responsible for managing my health condition.”

Read Part 2: Eliza and the Vulnerability Index

Read Part 3: NQF and the Standards for Patient-Reported Outcomes