PART I: THE PATH TO VARIATION REDUCTION
Chapter 1: Unwarranted Variation
In 1967, Jack Wennberg, then a young physician working in Vermont under a Medicare-sponsored grant, noticed distinct geographic variations in surgical procedures performed in the hospitals of New England. The rates of operations at one hospital versus another appeared to be medically unnecessary; that is, they seemed to have more to do with how many surgeons were at a particular hospital, and where they had trained, than whether a particular area had a higher incidence of an illness or condition. Dr. Wennberg called this phenomenon âunwarranted variationââhealthcare whose delivery could not be explained by illness, medical need, or the dictates of evidence-based medicine.
In one small town in Vermont, for example, surgeons had removed the tonsils of 70% of children 15 and under; in another town just two hours away, the percentage was only 7%.5 Intuitively feeling that 70% of children did not have diseased tonsils, Wennberg reviewed the data with the physicians who were removing tonsils at that elevated rate. They were surprised to learn of their collective overutilization.6 When Wennberg asked them to begin seeking second opinions on the need for tonsillectomies, they agreed, and over a five-year period, their common attention to the issue enabled them to reduce tonsillectomy rates by two-thirds, eliminating unnecessary medical procedures and creating significant cost savings for patients.
Preference versus Supply
As he continued his research, Wennberg began to examine Medicare expenditures for other types of procedures. (Because Medicare has kept records since its inception in 1965, itâs an invaluable source of data for researchers who want to understand the how, how much, and to whom of healthcare in the United States.) He found that he could classify Medicare expenditures into three categories:
- What he called effective care, grounded in evidence-based medicine, comprised only about 12% of all Medicare expenditures.
- Preference-sensitive care, driven mainly by the patientâs desire to obtain a certain test or procedure, amounted to 25% of Medicare expenditures.
- Supply-sensitive care, delivered at the discretion of the physician, accounted for 63% of all Medicare expenditures. This largest bucket of care was based not on medical evidence, but on expert opinion, and because expertsâ opinions vary, physicians could find an expert opinion to justify their varying decisions to test, prescribe, or operate.
When Wennberg found unwarranted variation in the small amount of evidence-based care, it was generally in the form of under-utilization of procedures; that is, physicians did less than they should have in caring for their patients. (And when unwarranted variation occurs as under-utilization, the cost is often measured in lives, not money. A recent example of under-utilization is physiciansâ failure to use beta blockers after an acute myocardial infarction. While the evidence is clear in the literature that using beta blockers saves lives, the utilization nationwide of this life-saving regimen is only about 60%.7)
But in the large majority of cases (preference-sensitive and supply-sensitive care), the pattern of variation was towards over-utilization. Because this book concerns those two latter categories, my focus is on over-utilization of medical procedures.
Wennberg submitted his findings to the prominent medical journals of the day, but encountered multiple obstacles to publication: most editors could not believe that individual physician practices could vary so much just on the basis of personal preference of physicians. When his work was finally published in 1973, it appeared in Science, a journal that typically publishes basic research, not clinical studies, and was barely noticed by the medical community.8
Warranted versus Unwarranted
Indeed, it remains difficult today for many physicians to believe in unwarranted clinical variation, even when shown the data. Itâs all too easy to think of other, more palatable, warranted reasons for variation in patient care, such as:
- The data are wrong.
- The data are correct but anomalous.
- My patients are sicker than other groups of patients.
- My medical group provides better service.
- We use different codes for that procedure or diagnosis.
And, in fact, some variation in treatment is warranted. Some physicians take care of patients with higher levels of acuity; or they are referred more patients for procedures because of an ability to perform a particular procedure that no one else in the area has the capability to do; or the population the physicians serve is truly different from the norm because of ethnic or socioeconomic characteristics. But unwarranted variation occurred in 1967 and it occurs through the present day.
While problems with unwarranted variation have been identified for nearly 50 years, little systematic success has been achieved in reducing that variation. Part of the problem may be that when discussing this issue thereâs no standard term used in either the medical or popular literature. The Institute for Medicine calls this issue âwasteâ and the Institute for Healthcare Improvement (IHI) calls it âoveruse.â9 The National Quality Forum speaks of âmisuse,â10 while noted health economist Victor Fuchs calls it ânot cost-effective.â11 Opponents to change speak of ârationingâ and âdeath panels.â12 These later characterizations are not inventions of Sarah Palin and her ilk. As a teenager in the fall of 1962, I recall reading a Life magazine article on a âlife and death committeeâ for allotting dialysis slots. As a medical student in the early 1970s, I remember discussing which of our patients would go on chronic dialysis, because even at that time a limited number of dialysis machines were available. Physicians have always agonized over these decisions, weighing the risks and benefits of every procedure and prescription. Politicizing the decisions does not make our choices any easier. Yes: sometimes our decisions are a matter of life and death, but thatâs part of what physicians do.
Value
The difference between variation thatâs warranted and variation thatâs unwarranted is a difference in value. Traditionally, value in healthcare treatment is defined as the quality of the service (which is made up of clinical outcomes, ready availability of the service, and patient satisfaction) divided by the cost of the service:13
or, more simply,
But, as Wennbergâs work shows, thereâs an additional factor thatâs impact is enormous: the appropriateness of the procedure.14 A physician can perform the highest-quality procedure at the lowest cost, for example, but if itâs the wrong procedure, or done on the wrong patient, its value is negative. Consider the high rate of removal of tonsils in some Vermont children. Everything in a given surgery may have gone flawlessly and quickly, the surgery may have been scheduled for the time and in the location the parents wanted, and the parents may have been highly satisfied with the surgeon, hospital, and outcome, but if the childâs tonsils didnât really need to be removed, the overall outcome was poor. So the value equation actually looks like this:
Value must be the focus and the measure for improvement in healthcare. It doesnât matter if weâre talking about improvement in clinical outcomes or improvement in how much we spend, value must be our main concern. Reducing cost without considering its effect on quality of care is folly, while a change that increases cost but also increases quality is something we intuitively know is valuable. The question is: how can we afford to make these improvements when the cost of healthcare is already so high?
Affording the New
I suggest that the only way weâll be able to afford the new and improved is to eliminate the inappropriate and unnecessary. I am not suggesting that new is always better than old. On the contrary, some older medications, tests, and procedures are still extremely useful. Aspirin has been around for more than 100 years and is still an effective and inexpensive anti-inflammatory medication, yet many doctors prescribe brand-name non-steroidal anti-inflammatory agents when generic aspirin will do. A simple test such as a chest X-ray can still diagnose many pulmonary problems, yet many physicians order the more costly and higher radiation of a CT scan even before obtaining the plain chest X-ray. Traditional surgery under local anesthesia is still an effective way to repair inguinal hernias, yet some surgeons now repair them through a laproscope, which requires general anesthesia. Many hospitals are adding robotic surgery units in an arms-race type competition with neighboring hospitals, even though there is little evidence that robotic surgery improves outcomes or decreases complications.
So before we choose to substitute new for old, we need to be clear that the change brings a gain in value. Does the benefit to this patient significantly outweigh the risk? If it does, then the value of the medication, test, or surgery is warranted; that is, itâs high enough to make it a necessary and appropriate part of this patientâs healthcare. Warranted, necessary, and appropriate are the keys to providing the care we all want, at a cost that we all can afford.
The Dartmouth Atlas
After leaving Vermont, Jack Wennberg joined the faculty of Dartmouth School of Medicine, where he founded the Center for Evaluative Clinical Sciences and expanded his research into variation to the national level. One of his achievements is the Dartmouth Atlas of Health Care (www.dartmouthatlas.org), a series of colorful maps of the United States that use increasingly saturated hues to display low-, medium-, and high-utilizing areas of various Medicare services. Whether the data represented are for hospital use, surgical procedures, post-acute care, or any other of a number of topics, map after map shows the same vast, unwarranted geographical variations in care.
One of the most intriguing of the Dartmouth Atlas maps (see Figure 1) shows Medicare expenditures in the last six months of life. This map shows the same kind of geographic variations that all the others do and is quite convincing even to those who are skeptical about the reality of unwarranted variation in healthcare: because these patients all had the same outcome (they died), it would be hard to insist that costs were higher in some areas because some p...