Ian Hargaves, PhD; Rene Rodriguez-Gutierrez, MD; Victor M. Montori, MD, MSC
Recently, while visiting a primary care clinic, one of the authors saw a sign posted on an exam room wall encouraging patients with diabetes to have their cholesterol measured. Amidst the noise of notices to be found in clinical spaces this seemed innocuous enough. The sign went on to explain that patients were encouraged to measure their low-density lipoprotein cholesterol (LDL-C) levels, so that the clinic could meet its quality target of 100% of diabetes patients with measured cholesterol levels. We also recently heard the story of a woman who underwent mammography, only because she did not want to affect her clinician’s screening numbers. In both cases, the rationale for therapy was cast in terms of meeting target quality measures, rather than in terms of doing what is best for the patient. In the care of patients with diabetes, a common marker of quality has been the achievement of tight glycemic control (eg glycated hemoglobin (A1C) - below 7%).1,2
Failure to respond to higher A1C levels with treatment intensification has been called “clinical inertia,” and patients who do not achieve this target are often seen as “noncompliant” or difficult.3
The focus on A1C is so pervasive, that a survey of patients, with diabetes, identified lowering A1C as a more important justification to try a new diabetes drug than avoiding amputations, blindness, or kidney damage.4
The only outcome surveyed patients ranked higher than A1C was avoiding death. How is it that lowering A1C, as a goal, can be second only to avoiding death?
We find ourselves in a situation in which quality targets may not be helping, and in some cases, may be harming patients struggling with the hard work of living with diabetes.5
Consider the case of Maria, a single mother of 3 who works 2 part-time jobs while living with type 2 diabetes and hypertension. Over the last 2 years, she has partnered with her primary care clinician to improve her diabetes control. During this time, she has worked on improving her diet and regularly taking 2 oral antidiabetic drugs, and a daily injection of long-acting insulin. Her A1C has dropped from 14% to 8%, she has lost weight, and she has been free of hypoglycemia. This is an achievement that Maria and her clinician consider worthy of celebration. Yet, in order to meet a reimbursement-tied A1C quality target <7.5%, she would likely need to switch to a self-managed and complex multi-dose insulin program. This approach is both demanding and expensive. This program will add a financial and treatment burden that may compete with, and even compromise, her ability to maintain the positive lifestyle changes that she has achieved, while contributing to weight gain and hypoglycemia. Is achieving the A1C target really the best for this patient? Was the clinic really not providing high quality care when her A1C dropped from 14% to 8%? Furthermore, are these “wins” not particularly relevant for practices that serve populations with difficult living conditions and poor health profiles, in which the resources to implement complex treatments are more limited? Perversely, by tying reimbursement to strict A1C targets, payers may be penalizing practices that are serving populations, most challenged by the demands of living with diabetes, and rewarding practices that actively exclude those patients from care.
HOW DID WE GET HERE?
Diabetes care is a practice that is rampant with measures—it is through these measures that we commonly detect diabetes, in the first place, and monitor the progression of a disease that often remains asymptomatic. The Diabetes Quality Improvement Project (DQIP), in the early 2000s, proposed A1C, lipid testing, LDL ≤ 130, blood pressure control, foot, eye and renal examinations as measures of diabetes care quality. These measures were widely adopted along with the Healthcare Effectiveness Data and Information Set (HEDIS),6
and reflected what was easier to retrieve and report to payers or the public from medical records and laboratory results. They do not, however, take into account how care comes together to advance the situation of a given patient.
Many diabetes patients present with multiple competing conditions, many of which are also chronic.7
Being chronic, these conditions invariably intertwine with one another and the economic, time, familial, and social burdens of day-to-day life. For these patients, it is harder to isolate diabetes as a discrete illness or to separate its management from the other demands of life. Early, single-disease-specific measures, that offer a narrow measure of quality such as A1C, neglect the quality and character of the work that patients and clinicians are doing in treating lives lived with diabetes.
GOALS AND WHAT IS BEST FOR THE PATIENT
Quality targets work well when they promote a practice that is effective, safe, and feasible, regardless of the presence of comorbidities and social and economic complications. In short, quality targets work well to drive compliance when medical science clearly knows what is best for the patient. Most diabetes care fails this criterion. Lowering A1C to <7.5%, for example, is associated with modest beneficial effects, at best, for the average patient.8
In Maria’s case, it is difficult, and likely futile, to isolate diabetes as an object of care given the complex tapestry of personal, social, and economic concerns that characterize her situation. Achieving levels of A1C <7.5% is clearly possible, yet it is not clear that in achieving this target we are providing high-quality care for Maria.
Contemporary diabetes care is not a practice of clearly knowing what is best and applying that knowledge to achieve high-quality care. Diabetes care is practiced in conditions in which it is far from clear what the best course of action is for the individual patient and her family. Goals such as quality targets that dictate action in situations where we clearly know what is best for a patient and her family, do not necessarily work in situations in which we remain uncertain as to what “best” is.9
In the context of the intellectually, practically, and emotionally troubling situations of life with comorbid diabetes, medicine must go beyond the application of knowledge. It must partner with patients to create courses of action that address the specific challenges of each patient. When what is best is unclear, processes must be found to find a coherent way forward. These may include activities such as shared decision making—a deliberative act in which patients and clinicians think and talk through hypotheses of how to proceed.10
In the course of these conversations, goals may emerge—for example, to work on strengthening the patient’s mental state and supportive relationships so that she can better deal with challenges. These situational goals, however, will have different qualities than existing fixed targets, specifically:
1. Situational goals arise in order to attend to the problem of an individual’s situation. In so doing, a goal will help to discern, from the tangle of contextual complications, the nature of the particular problem that currently requires action, along with the means by which the problem may possibly be addressed.
2. Because the function of a situational goal is to help clinicians deal with the problem of a patient’s particular situation, it is integrally connected to the problem at hand. In contrast, external arbiters apply fixed quality targets without regard for their applicability to the problems faced by each patient.
3. Situational goals are flexible and respond to changes in circumstances—as the problems of life with diabetes change, patients and clinicians will develop changing goals to attend to new circumstances.
Fixed quality targets are not intended to attend to the problems of living with and treating diabetes. More commonly, they are adapted to problems of policy, organizational management, and safety. There are significant challenges in all of these aspects of providing diabetes care, and fixed quality targets undoubtedly have a role to play in ensuring that best practices are followed. We should, however, be mindful of the problems that targets were designed to address and modify them when they no longer serve. In some cases, targets like, A1C, belong to a time when the problem of getting a quality measure implemented was more important than getting the right measure implemented.
Arguably, goals and targets are not developed for clinicians and patients to achieve them, but to draw attention to problems in treating and living with diabetes, and open consideration of appropriate ways to address them, given each patient’s situation. Nothing is more frustrating for clinicians than to feel pushed to do the wrong thing for a patient by a misguided quality target. If quality targets are designed to improve our practice, then these targets should promote the best practices of patient-clinician collaboration to address the problems of life with comorbid diabetes. This may include, as in the case of Maria, knowing when to celebrate. We may then find that quality in diabetes care is not the application of what medical science knows to be best. Rather, it is finding kind and careful ways forward when what is best is far from clear.
For the last decade, the Patient Advisory Group, a group of patients with diabetes from the community, has met with investigators of the KER Unit, at Mayo Clinic, to ground their work on what matters to patients. The insights developed here would not have been possible without their generous contribution to the science of healthcare.