As we grapple with provider shortages, the surge in chronic illness and the quality to price (QPR as they say in the wine business) challenge in US healthcare delivery, it’s hard to imagine a future that does not include some sort of guideline or algorithm-driven care. As providers take on more financial risk, one common strategy involves team-based care, and the attendant increase in decision-making and care delivery by non-physician clinicians. If the je ne sais quoi feature of a quintessentially great doctor is clinical judgment and instinct, one of the challenges of this transition to team-based care is how to harness that trait and use it efficiently.
Care decisions that are unassailable at a population level (e.g., women should have regular, routine PAP smears or smoking is bad for your health) or are algorithmic in nature (e.g., titration of treatment for uncomplicated hypertension or therapy for mild to moderate teenage acne) can all be effectively reduced to guidelines. This, in turn, allows a physician to delegate certain therapeutic decisions to non-physician providers while maintaining a high degree of care quality. It is also thought that this type of uniformity of care delivery will improve the QPR too, by decreasing variability.
How do we come up with guidelines? Typically they are based on large-scale, randomized, controlled clinical studies. As is nicely articulated in a recent JAMA opinion piece by Drs. Jeffrey Goldberg and Alfred Buxton (JAMA, June 26, 2013—Vol 309, No. 24, pg 2559), guidelines are formulated based on the inclusion criteria for these trials. This process gives us comfort that guidelines are based on rigorous science — and that is a good thing. The challenge arises when we realize that individuals do not reflect populations exactly. Clinical research is much more complex than wet lab work because people are complex and indeed unique. Every clinician has had the experience of prescribing a therapy to a patient who fit guideline criteria exactly and having the opposite outcome of what the guideline predicts.
Goldberg and Buxton point out the collision of this guideline-based care delivery model with the burgeoning area of personalized medicine. I was immediately drawn to their definition of personalized medicine: “The tailoring of medical treatment to the individual characteristics of each patient. It does not literally mean the creation of drugs or medical devices that are unique to a patient, but rather the ability to classify individuals into subpopulations that differ in their susceptibility to a particular disease or their response to a specific treatment.” I always felt like there was too much emphasis on the genetic components of personalized medicine.
Our vision at the Center for Connected Health (which is backed up by our experience to date) is that we will get far richer and complex data from multiple phenotypic inputs such as physiologic monitoring data, mood and motivation-related data than is represented by genomic data. The genome is an incredibly important anchor for devising a personalized medicine profile, but the profile will change over an individual’s lifetime according to these phenotypic inputs.
We’ve done some preliminary work on this and found that indeed we can map individuals phenotypic data over time as they go through an intervention designed, for example, to improve activity level. During a six month period of tracking activity and motivation, we have seen dynamic changes in these two variables. Think about it over a lifetime.
The collision with guidelines is multifactorial. We are all individuals and none of us are completely representative of the composite patient who is defined by the inclusion criteria for the clinical trial that lead to the guideline. Thus, some of us are bound to be poor candidates for the prescribed intervention (I hate to mention it, but we’ve all seen examples of Uncle Harry who smoked two packs per day, lived into his 90s and died of causes unrelated to smoking). If that wasn’t enough, there is the fact that we change over time and though we might fit a guideline today, we may not in a year.
Really, when you think about it, ‘clinical judgment and instinct’ is the 20th century (and earlier) embodiment of personalized medicine. Those of us who are clinicians can all point to experiences where we’ve said, “I can’t tell you why, but I really think we should do it this way” (this way being contrary to conventional wisdom) and it has generated a positive outcome. Of course we also have experiences where the outcome is not good or where we make mistakes that could have been prevented by adherence to guidelines.
How to make sense of this complex and contradictory situation? Here’s my take:
- Personalized medicine, however you define it, is still in the very early stages. We have decades to go, probably on both the genetic and phenotypic fronts, before we can comfortably replace guidelines.
- We should welcome the sharing of decision-making across the care team and maximize the use of non-physician clinicians. Guidelines give us the state-of-the-art way to do this.
- The best form of personalized medicine today is still clinician instinct and judgment. This does not mean deferring all clinical decisions to the most senior or most highly trained person on the team. The care delivery culture can be modified to maximize appropriate personalization of care while adhering appropriately to guidelines. This requires an open culture where inquiry is encouraged. Each care team member must be comfortable with what he or she doesn’t know, with spotting exceptions to norms and engaging other team members in a learning dialogue around these exceptions.
This should enable guidelines to be appropriately applied while surfacing exceptions for discussion. In the meantime, we and others will be working as fast as we can to create the framework for personalized medicine from both the genetic and phenotypic perspective.