Traditional industrial QI does not work in primary care. A recent publication brilliantly makes this case on both theoretical and practical levels ;-).
On a theoretical level, primary care is best thought of as a complex adaptive system, not a simple linear mechanical system. Think of complex adaptive systems as kind of being like a series of tipping points with lots of players in the environment. Relationships between the players are complex. Big changes can occur with small perturbations of the system and large perturbations can cause little change.
Knee replacement surgery is a great example of a linear mechanical system. Antibiotics should always be given before the surgery, the surgery should always occur in the same sequence, and post-op blood clot prevention should always be ordered. The patient today, the patient tomorrow, the patient in another state, all will benefit from the same service provided the same way each time. In this case, traditional QI approaches such as PDSA cycles are totally appropriate and produce similar improvements in outcomes in the same way the traditional QI approach improves the quality of a Toyota product. How can we insure that the patient always gets her antibiotic prior to incision? Hard wire that step into the process.
In contrast, if a family physician sees a patient in his clinic with an elevated blood sugar level and advises the patient to start taking insulin, what will happen next? Anything. The next time the physician sees the patient the sugar could be anything from dangerously low to even higher than before. The same clinical situation in a different patient seen the same day could result in a wildly different outcome. The same patient having this step repeated a year later could have a completely different outcome. The reasons for this are the myriad forces all family physicians understand: dozens of social determinant barriers, co-morbidities, quirky patient beliefs, and so on. These are example of non-linear complex processes.
The authors give examples of how care processes are different between linear mechanical and non-linear complex processes. They are different by the complexity of the process: linear processes are simpler with fewer variables to control that are also easier to measure. They are different by how standardized a process should be: in linear processes antibiotics should always be given before an elective major surgical case. Complex processes should not be standardized: chest pain in a healthy 18-year old should be managed differently than chest pain in a 72-year-old diabetic hypertensive smoker. They are different in how the processes are controlled: a linear process often involves a human who has essentially been turned into a machine by virtue of being rendered unconscious. Complex processes are driven by a milieu of forces including unique patient beliefs and priorities, socioeconomic forces, and the external environment.
The authors demonstrate that the outcome goals are different between mechanical and complex processes. In a linear process, the goals are not influenced by co-morbidities or other factors: the hip replacement surgery happens the same way whether the patient has 5 chronic diseases or none. In a complex process, one patient with metastatic cancer might choose another round of toxic chemotherapy while another chooses hospice, and yet another chooses an unproven alternative treatment. In a linear process, the goals are also clear in that the doctor, hospital, and patient all agree on the definition of a successful outcome: a patient who is successfully extubated from mechanical ventilation. In a complex process, the goals may be vastly different between doctor and patient: a patient with high blood sugars declines to start recommended insulin to reduce her blood glucose level because of concerns about the affordability of the medicine and a belief that insulin killed her favorite aunt. In a linear process, the timing of the outcome goal is not in question: every time an elective surgery occurs a list of interventions should stick to a fairly rigid schedule. In a complex process, the timing of a desired goal may be negotiated to come at a nebulous date in the future: the patient who just lost her husband will restart her new exercise program at some date in the future when she doesn’t feel like all the energy has been drained from her body, if ever.
For the last examples, consider this comparison. Doctor A works in a upper middle income neighborhood where every patient has health insurance and there are 2 Whole Foods within 2 miles of his clinic building. His diabetic patients’ average hemoglobin A1c is 7.4. Doctor B works in a Federally Qualified Health Center. All of his patients are poor. Many are undocumented immigrants. His diabetic patients’ average hemoglobin A1c is 8.8. Who reading this believes that these data mean that Doctor A is a higher quality doctor than Doctor B? No one with any sense.
The authors show how the current obsession with quality scorecards are wholly inadequate for family medicine. First, there are no adequate risk-adjustment tools to correct for the previous example. Studies have shown that the same doctors doing the same work in a private well-insured hospital and a safety net hospital are ranked as being high quality doctors in the former and poor quality doctors in the latter. Second, the existing six sigma mentality transferred to healthcare by the early proponents of QI/PDSA means that the numerical target should always be to strive to achieve the stated goal 99.99966% of the time. This is a ludicrous target for everything in primary care. Third, quality scorecards inadequately reflect the breadth and depth of family medicine. If an orthopod only does hips and knees, then a scorecard of his hip and knee infection and blood clot rates mostly reflect the quality of his work. If a family physician is scored on even 20 different measures, that leaves off the report approximately 600 other symptoms and diseases she manages. This is also ludicrous.
In the current belief system of the pundit/political class, they want to “pay for value, not volume.” In Medicare’s current Common Core Measures for Primary Care, they have come up with mostly a list of these simplistic single-disease metrics that feed into a “value-based” scorecard. Nothing could be farther from the truth.
All they are about to do is to incentivize family physicians to dump the most vulnerable patients from their roles. The poor patients, those with mental illness, difficult family situations, and a history of poor life choices will be cast out of the primary care system and thrown back to the revolving ER door system: the worst place these patients can be. One gets what one pays for, and Medicare is about to pay for this horror.
A high-quality family physician is not one who mindlessly follows simplistic single-disease recipes. In my world, we call this practicing like a nurse practitioner. A high-quality family physician often achieves the best care by NOT doing what the cookbook says. In contrast to the horror Medicare is about to foist on the American people, a high-quality family physician says, “I will take care of you no matter what. It doesn’t matter your background, skin color, social status, mental fitness, education level, or quirky beliefs. My door is always open. We can even disagree on what we think the best path is for you, but I will always respect your wishes (except for the rare times where my calling demands that I really need to protect you from yourself). We will walk together on your life’s journey and I will never abandon you.”
That’s what a great family doctor sounds like, and no simple-minded mechanical linear report card will ever figure this out.