The US health care system frequently focuses on reducing “unnecessary care variation” and with good reason. Recent literature places the cost of variability from best practices in care in the billions, with one review of nearly 200 studies demonstrating unnecessary variation in all specialties and across the full spectrum of clinical outcomes and care environments. As physician leaders, we too have each spent significant energy driving transformation to minimize variability within health systems and have seen these efforts lead to higher quality, lower total cost of care, and better patient experience firsthand.
But clinicians appreciate that reducing variation is only part of the story. The truth is that one size may fit most but rarely fits all. To deliver truly patient-centered care, we must both reduce unnecessary practice variation in care and increase truly necessary care variation. Unfortunately, today’s care transformation efforts largely ignore this second, critical part of both care model design and outcome measurement. As a result, most approaches lack the ability to define, measure, understand, and appreciate necessary variation in care, and thus fall short of achieving true clinical excellence for our patients and professionals.
Defining Unnecessary Care Variation
How do we simultaneously drive down unnecessary variation and drive-up necessary variation to achieve optimal outcomes? First, we must define what we mean by necessary variation. Unnecessary variation flows from physicians and the teams and processes available to support us in our practice. This includes differences in when, how, and where we were trained (such as diuretic use in neonatology), our practice environment (for example, a large academic trauma center versus a small critical access hospital), our propensity to fall victim to heuristics such as the availability bias or anchoring bias, and even single-patient, single-episode encounters that change a physician’s treatment decisions for an entire career. In contrast, necessary variation flows from characteristics unique to an individual patient.
Evidence matters, but perfect evidence rarely exists, and it is never the only thing that matters. The spectrum of clinical and sociodemographic factors embodied in real-life patients are poorly captured by a clinical trial’s “median” patient. Physicians readily acknowledge that no evidence defines the right thing to do for every single patient (for example, beta blockers do not always work for heart failure nor dexamethasone for premature infants in respiratory failure). Shared decision making also drives necessary variation, as making real care decisions involves balancing available evidence with different attitudes toward risk, prior experiences, religious, or other preferences and fears (whether rational or otherwise) for both patient and provider.
Most care metrics fail to capture the myriad social determinants that drive health and care. Solving for these very real factors can require clinicians to modify standard care practices, such as accounting for homelessness in post-op discharge planning in the same way we account for penicillin allergy in pre-op antibiotic choice. This real-world tapestry leads the authors to define necessary variation as variation from an otherwise standardized care process required to support an individual patient in reaching his or her personally optimized outcome.
Measuring Necessary Care Variation
So, how can we tell if the “variant doctor” (and care team) is really personalizing care to support a patient in reaching their personally optimized outcome? When does “going off book” represent the best and the right thing for the patient, and when is it unnecessary variation that should be eliminated? When is driving alignment around standard, evidence-based best practice a laudable effort to reduce care variation, and when does it hinder our professional obligation to personalize treatment to match a patient’s unique pathology and preferences? Is even trying to solve this challenge a fool’s errand?
We don’t think it is. We propose the standard implementation of balancing measures for care transformation efforts that both identify and quantify aspects of necessary, valuable variation. These measures must capture both how easily and frequently clinicians execute an agreed-upon best practice and capture what happens when a clinician decides to vary from the best practice pathway.
Measurement capabilities exist in today’s analytics environments and have been there for more than a decade—we just have to decide to use them. One published example hardwired a 40-element best practice pathway for coronary artery bypass surgery into an electronic health record (EHR) in a manner that captured necessary variation by design. The EHR workflows to execute the pathway required that clinicians either comply with best practice or document the rationale for not using a specific best practice element—a step designed to protect physician prerogative and provide ongoing feedback on guideline appropriateness.
Beyond individual circumstances with well-defined best practices, there are numerous clinical scenarios where best practices either don’t exist or represent best historical understanding and anecdotes rather than evidence. The dominant method for creating clinical knowledge, the randomized controlled trial (RCT), is not typically used to address questions directly relevant to common practice settings. Instead, RCTs have been and will continue to be used to test for treatment benefits in highly selected populations with a low comorbid disease burden. As a result, there is an opportunity to carefully analyze and take advantage of necessary variation when clinicians are required to fill in gaps where either they lack knowledge or where no knowledge yet exists. In those circumstances, analytical review of EHR data may enable the creation of real-world evidence relevant to everyday practice needs by facilitating real-time use of knowledge in practice.
Once we begin systematically capturing and measuring necessary variation, we must design care transformation processes to understand it. Sometimes variance from a care pathway educates care transformation work and improves the evidence-based pathway. Other times, such variation may represent a learning opportunity or moment of accountability for the clinician. WakeMed has recently started analyzing pathway adherence in tandem with health equity, offering an opportunity to both uncover potential for bias and ensure that care transformation initiatives are equitably raising the bar across the age, race, ethnicity, sex, and geographic spectrum.
Meeting Patient Needs
In particular, we are most interested in a fourth option—specifically, when variation from a best practice represents a provider practicing at the top of her license, bending a standard treatment course to appropriately meet a patient’s unique needs. Whether the justification is a patient’s unique pathology, unique preference, or unique care circumstances, these variations must be identified, celebrated, and reinforced as core to our professionalism.
In doing so, we uncover the previously ignored magic in balancing the yin and yang of necessary and unnecessary variation in care. Celebrating necessary variation will help physicians concede that best practice pathways are not a tool to crush autonomy, but rather a guideline that makes good sense for most patients most of the time.
Imagine you’re on-service and your team has fully embraced the yin and yang of care transformation. You move so efficiently through most of your rounds that you get to spend extra time and energy with the small but critical subset of your patients who can most benefit from your expertise and still finish faster and more professionally fulfilled. By more efficiently accomplishing “most” of what we do (decrease unnecessary variation), we open just that creative space needed to meet unique patients’ unique needs with more creative energy (increase necessary variation). Such energy is both patient-centric and professionally renewing, and we believe it can drive a better, more sustainable flywheel of continuous improvement than focusing only on “waste.”
In the end, physicians must own both sides of our professionalism. Just as we would consider failing to meet a patient’s evidence-based care needs as subpar, so should we concede that eliminating necessary care variation similarly fails to meet medicine’s professional demands for optimal patient care.
Can we have our cake and eat it, too? We think we can.