Tracking Medicine: A Researcher's Quest to Understand Health CareBuy the book!
The Author Excerpt Reviews Resources Contact Home  

Excerpt from Tracking Medicine

In Health Care, Geography is Destiny
Early in my career, I was hired as director of a federally sponsored program whose goal was to ensure that all Vermonters had access to recent advances in the treatment of heart disease, cancers, and stroke. As part of the program, my colleagues and I developed a data system that we thought would help us identify which Vermont communities were underserved, and thus in need of the program’s help. As the results came in, however, rather than evidence for underuse (i.e., patients not getting care they needed), we found extensive and seemingly inexplicable variation in the way health care was delivered from one Vermont community to another. In Stowe, for example, the rate of tonsillectomy was such that by age 15, about 60% of children were without tonsils, while in the bordering town of Waterbury, only 20% had undergone the surgery by that age. Among communities, the chances that a woman would have her uterus surgically removed varied by more than fourfold, and the rate of gallbladder surgery varied by more than threefold. Rates of hospitalizations for a host of different medical conditions also varied in ways that made little sense; on a per capita basis, patients were hospitalized in Randolph two times more often for digestive disease than in Middlebury and three times more often for respiratory disease.

These are just a few examples of the chaotic patterns of utilization and practice our data uncovered – variations that challenged the very premise of the program I had been hired to direct. The rates of hospitalization and surgery appeared to be unrelated to illness or other patient-based factors, and thus the variation was at odds with the conventional wisdom that medicine was driven by science and by an understanding of patient desires and preferences. The data also challenged the assumption that the supply of medical resources and the capacity of the health care system were regulated either by a central professional consensus on the need for medical care, and its effectiveness, or by the invisible hand of the market. It became clear that the amount of care Vermonters received depended on where they lived and on the physicians and hospitals they used.

Over the years since that time, my colleagues and I have pursued the study of practice variation in many places, using a variety of methods, and the Vermont findings have been widely confirmed. Unwarranted variation in health care delivery – variation that cannot be explained on the basis of illness, medical evidence, or patient preference – is ubiquitous. Moreover, as I argue in this book, an understanding of the causes of unwarranted variation has important and sometimes surprising implications for today’s debate over health care reform. Most analysts of health care reform expect huge increases in spending once the uninsured gain coverage and begin to consume more health care services. But the understanding I have gained from the study of the practice variation phenomenon provides a counterintuitive, maybe even shocking, prediction: given the important role that the supply of resources plays in determining utilization of medical care, increasing the insured population will have a much smaller impact on the trend in overall health care costs than estimated, provided that the capacity of the health care system is not increased.

Another prediction that emerges from an understanding of practice variation is that controlling costs will not necessarily require rationing – if by “rationing” we mean the withholding of care that patients want, and that is effective in improving outcomes. The studies reviewed in this book show that much of health care is of questionable value and that informed patients often prefer a form of treatment other than the one their physicians actually prescribe. Indeed, when offered a clear explanation of the treatment options, informed patients often choose the less invasive treatment, resulting in a decline in the use of elective surgery and certain cancer screening tests. Moreover, more care is not necessarily better, at least when it comes to managing chronic illness. Care coordination and intelligent management of patients over the course of their illness, which typically lasts until death, count far more than simply providing more medical services. Some of our most respected health care providers – for example, the Mayo Clinic, the Geisinger Clinic, and the Cleveland Clinic – provide high-quality care at a much lower per capita cost than most other providers. If the rest of the nation were equally efficient, we could shave 30% to 40% off the cost of caring for Medicare’s chronically ill patients.

If, as I recommend in this book, health care reform concentrates on four goals, the quality and value of care will increase and growth in health care costs will likely decrease. Those goals are as follows:

  1. Promoting organized systems of health care delivery;
  2. Establishing informed patient choice as the ethical and legal standard for decisions surrounding elective surgeries, drugs, tests, and procedures, and care at the end of life;
  3. Improving the science of health care delivery;
  4. Constraining undisciplined growth in health care capacity and spending.

These are strong conclusions, ones that policy makers should not ignore given today’s economic realities. They are supported by a growing body of evidence drawn from practice variation studies and from interventions to improve the scientific basis of clinical decision making and promote informed patient choice. An important goal of this book is to make this complicated and interconnected body of research accessible to a broad audience, including policy makers, health care providers, students, patient advocates, and, I hope, patients and families.

Surgical Variation: Understanding Preference-Sensitive Care
Patients traditionally delegate decision making about treatments to their physicians, under the assumption that physicians prescribe treatments based not only on medical science but also on an understanding of what is best for the individual patient. Embedded in the idea that the physician knows what is best for the patient is the notion that the physician also knows what the patient wants. Yet as early as the 1930s, it was evident that local medical opinion was behind the marked variation in tonsillectomy rates, rather than clinical science or patient (parental) preference. By the mid-1970s, it became clear that the rates of utilization of most common surgical procedures varied extensively among regions, some more than others. Physicians everywhere seemed to differ among themselves on the value of many operations, on who would benefit and who would not, and these differences of opinion directly influenced the incidence of any given surgery. What the patient wanted often appeared not to matter much.

The importance of scientific uncertainty and the misdiagnosis of patient preferences as a cause of practice variation became most clear over the course of a decade-long research effort we undertook to understand why surgery rates for an enlarged prostate showed such great variation among regions in the state of Maine. It turned out that while many urologists undertook surgery under the assumption that it prolonged life, our research showed this was not the case. However, surgery did have a potentially positive effect on the quality of life by reducing symptoms. But this benefit had to be weighed against the harmful effect of surgery on sexual function. It became evident through our research that rational choice required the active engagement of the patient in the choice of treatment. Similar research sponsored by a new federal agency, to investigate practice variations for back surgery and cardiac surgery, came to similar conclusions regarding the importance of patients understanding the benefits and harms of alternative treatments and making informed choices.

Establishing a market where the utilization of preference-sensitive treatments is determined by patient demand will require a profound cultural change in the doctor-patient relationship: replacing delegated decision making, and the doctrine of informed consent, with shared decision making and informed patient choice as the standard for determining the medical necessity of preference-sensitive treatments.

Remedying variation in preference-sensitive care will also require a change in the research culture, a topic that is relevant to today’s debate over evidence-based medicine. Under the Agency for Health Care Policy and Research (forerunner to the Agency for Healthcare Research and Quality), a collaborative research model was established that was highly successful in reducing scientific uncertainty and promoting informed patient choice. However, it also challenged the conventional wisdom; it was abruptly interrupted by the U.S. Congress in the mid-1990s, in large part because of strong negative reaction from surgeons over the conclusions of the back surgery research team.

Studies discussed in this section suggest that the implementation of shared decision making will reduce the utilization of surgery and save money. But addressing unwarranted variation in preference-sensitive care is essential not just for economic reasons. By failing to take into account patient preferences, and enabling patients to make informed choices, surgery will be misused. It should be considered a serious form of medical error when surgeons operate on patients who would not have wanted the procedure had they been fully informed and empowered to participate in a meaningful way in the choice of treatment.

Medical Variation: Understanding Supply-Sensitive Care
Supply-sensitive care differs in several fundamental ways from preference-sensitive surgical procedures. First, the physicians whose decisions determine the frequency of the use of supply-sensitive care are mostly primary care physicians and medical specialists, not surgeons. Second, supply-sensitive care is not about deciding on a specific treatment but rather about how frequently numerous, everyday medical services are provided in the process of managing acute and chronic illnesses. Supply-sensitive care covers a range of services, including physician visits, referrals to specialists, diagnostic tests, imaging exams, hospitalizations, and stays in intensive care units. Third, such decisions are not governed by strong medical theory, much less medical evidence. Indeed, practice guidelines governing clinical decisions involving supply-sensitive care are virtually nonexistent.

In the absence of evidence concerning effectiveness but under the prevailing cultural belief held by both patients and physicians that more health care is better, physicians use available capacity up to the point of its exhaustion. They schedule revisits up to the point where they have no time for more, they hospitalize patients until hospital beds become scarce, and they order more imaging exams whenever imaging equipment is available. In other words, as illustrated in Chapter 8, clinical decision making on the part of physicians determining the use of such care is sensitive to the level of supply of the resources available in a region.

This accommodation to supply seems to occur without awareness on the part of physicians that per capita supply varies from place to place or that it influences their behavior. As I will show, physicians practicing in Boston and New Haven (and affiliated with some of the nation’s most prestigious academic medical centers) were largely unaware that bed capacity (and hospitalization rates for supply-sensitive conditions) in Boston was 60% greater than that in New Haven on a per capita basis. Moreover, decisions that led to an expansion in capacity in Boston were made without information on baseline supply or what the expected benefits of an increase in supply might be.

Not surprisingly, the idea that the supply of resources exerts such a strong effect on utilization has not been met with open arms by the medical community. The major counter-argument against our interpretation of supply-sensitive care is that illness is the cause of the variation. Chapter 9 guides the reader through our studies, which show that differences among regions in the prevalence of chronic illness do not explain variation in the frequency of physician visits, hospitalizations, and other forms of supply-sensitive care. The same is so for demographic factors such as age, sex, and race. While black patients with chronic illness tend to get more supply-sensitive care than other racial groups, black patients living in regions with low overall use of supply-sensitive care receive less care than non-black patients living in high-use regions. The factor lurking behind these patterns of utilization is the capacity of the system – the number of hospital beds and physicians per capita.

The bottom line question is whether more is better. In the absence of clinical research, supply-sensitive care is perhaps best described as a black box, or mystery medicine. There is no corpus of scientific evidence that can be marshaled to answer the question of whether or when more care is better. Chapter 10 describes our epidemiologic research that looked into the marginal impact of increasing the intensity of supply-sensitive care on survival and on patient satisfaction. In both cases, we find evidence that, at least for Medicare enrollees in traditional fee-for-service medicine, more is not better; indeed, it may be worse. In other words, the problem is the overuse of care in high-intensity regions, not the rationing of care in low-intensity regions.

This is not merely a problem at community hospitals. Overuse also plagues patients who are cared for in the hospitals that are considered the nation’s best. But when it comes to managing chronic illness, it is not necessarily those who have the best reputation, including such recognition as high ratings from U.S.News & World Report, who are doing the best job. Chapter 10 examines the management practices of prestigious academic medical centers to show that, despite their reputations as bastions of medical science, their practice patterns show about as much variation as other hospitals across the United States. However, some hospitals are, in fact, better. Organized systems of care such as the Geisinger Clinic and Intermountain Healthcare are typically more efficient in the way they manage chronic illness, providing high-quality care at lower costs.

There are compelling reasons why the nation needs to challenge the way chronic illness is managed in the United States. Variation in the intensity of care, particularly the use of acute care hospitals, is the major cause of the more than two-fold variation in Medicare per capita spending among regions. It is not the prices, it is the use of care – the volume – that matters more. Given the lack of coordination between sectors of care, what logically seem to be sensible strategies for reducing inpatient care – building skilled nursing home beds or encouraging home health care, for example – simply do not pay off. The overuse of care by the chronically ill is getting worse everywhere but more so in regions that already are at the top in care intensity.

The problem is not just in Medicare; variation and overuse affect those under 65 years of age and appear to be highly correlated with variation in Medicare, which is not so surprising, given the importance of local capacity in influencing utilization. Reducing the volume (overuse) of care in high-use regions will benefit taxpayers and patients and families by reducing the subsidies from more efficient regions to help pay providers in less efficient regions and high co-payments by patients living in high-cost regions. It will also reduce the medicalization of death.

Pathways to Reform
My understanding of the sources of unwarranted variation and undisciplined growth in health care points to the importance of replacing disorganized, chaotic “systems” of care with organized systems; replacing delegated decision making and the doctrine of informed consent with shared decision making and the standard of informed patient choice; improving the science of health care delivery; and constraining undisciplined growth in health care capacity and spending.

The final section of my book sets forth some ideas and strategies for how these goals might be accomplished. The focus for Chapter 13 is on promoting the growth of organized care by providing economic incentives to providers who accept responsibility for caring for their population of loyal patients – not just in the acute phase but throughout the course of their illness, a strategy that seems particularly suited for chronically ill patients. The economic incentive is shared savings – the opportunity for providers who become more efficient to retain part of the savings to reinvest in care and reduction of debt. The shared savings strategy could result in large rewards for providers in high-cost regions who reduce their inpatient spending to the per capita levels of providers in low-cost regions. Academic medical centers are called on to undertake the necessary research to rationalize the clinical pathways for managing chronic illness – and to adjust their resource inputs toward the efficiency benchmarks that emerge from their research. Shared savings would be key to the completion of this mission.

Chapter 14 considers several approaches to promoting shared decision making and establishing informed patient choice, including changes in state laws governing malpractice to provide greater immunity to physicians who provide high-quality shared decision making, demonstration projects, economic incentives, and the assumption by primary care physicians of advocacy and professional accountability for ensuring informed patient choice as a standard of practice. Comparative effectiveness research along the lines discussed in Chapter 7 would make it feasible for primary care physicians to assume this role.

While the nation desperately needs to reengineer clinical practice, it cannot depend on reform of the delivery system as the primary means for controlling capacity and spending. In this upside-down economy, reform requires working from the top down as well as from the bottom up. In Chapter 15, I outline six steps that can be taken to place limits on capacity and spending and buy time for reform of the delivery system to take hold. The final chapter summarizes the challenge of practice variation.

The Challenge of Practice Variation
It is the thesis of this book that an understanding of unwarranted variation in health care delivery – variation that cannot be explained on the basis of prevalence of illness, medical evidence, or patient preference – provides a framework for interpreting the crisis in costs and the chaos that plagues health care in the United States. The chaotic patterns of practice, first uncovered in Vermont and New England, and now documented by the Dartmouth Atlas Project across the nation, are incompatible with the assumption that clinical science and medical ethics govern the utilization of care. While policy makers recognize that delegating choice to the seller of services sets up the possibility for abuse, most have assumed that undue influence on utilization by physicians was rare, the result of the action of an unscrupulous few who transgressed the dictates of clinical science or the standards of medical ethics. But supplier influence on utilization is not restricted to just a few bad apples – a handful of greedy physicians. It is a ubiquitous phenomenon, a central tendency of the market for health care services. It affects patients everywhere, including those cared for by the nation’s most famous academic medical centers, and by the most caring and careful of physicians.

Engaging Providers, the Federal Government and Payers
Practice variations pose a special challenge to medical science. The nation looks to academic medical centers and the National Institutes of Health (NIH) as the principal sources of medical innovation, and guarantors of the scientific basis of care. Yet the NIH and academic medicine have paid little attention to the science of health care delivery. The consequences of this neglect are evident in the inconsistent practice patterns of the academic medical centers themselves: the more than two-fold variation in the number of physicians and intensive care beds used in managing chronic illness, and the even greater variation in rates for preference-sensitive surgery.

Neglect of the sciences of health care delivery has other consequences. America’s medical schools do not teach the skills required to understand patient preferences, evaluate medical practice, assess clinical evidence, design and test clinical pathways, improve quality, and understand the effect of systems of care on clinical practice. The mentorship of physicians-in-training takes place in the context of widely varying practice patterns, such that those trained in Los Angeles and Miami teaching hospitals encounter an entirely different strategy for managing chronic illness than those trained in teaching hospitals in Salt Lake City, Portland, Oregon or Minneapolis. Surgical training programs routinely take place in settings where informed patient choice is not yet the standard of practice for establishing the need for elective procedures.

This must change. Over the past 50 years, the NIH has been remarkably successful in promoting the growth of biomedical sciences. This success is a testimony to the responsiveness of the nation’s medical schools and academic medical centers to the incentives embedded in an enlightened federal science policy. Now it is time for that science policy to provide similar incentives to establish the science of health care delivery. But it isn’t just a stable source of research funds that is needed. The results of this research can change medical theories and practice patterns, and thus lead to economic gain or loss to clinicians, hospitals, device makers and pharmaceutical companies, to name just a few of the players in the health care market. Anything that threatens their revenue stream is vulnerable to attack, as I learned from experience (see Chapter 7). To avoid a fate similar to that of the Agency for Health Policy and Research and the patient outcomes research teams, it is essential that the federal agency responsible for managing the new medical research agenda – for setting priorities, conducting peer review and awarding grants and contracts to researchers – be strong enough to protect its agenda and the scientific teams that conduct the research.

The success of any reform effort to address unwarranted variation will also depend on the emergence of professional leadership. On this point, I am quite optimistic. When confronted with practice variation, practicing physicians, nurses and other health professionals have stood up and done the right thing. I have told some of their stories in this book: the actions of the physicians in Morrisville, Vermont, that led to a rapid drop in the misuse of tonsillectomy in their own hospital; the campaign of Dan Hanley to establish feedback on variation in practice patterns to providers in Maine, and create the Maine Medical Assessment Foundation; the engagement of Maine’s urologists in a ten-year study of their own practice patterns, to name only a few.

The leaders of organized systems of care – private sector organizations such as the Mayo Clinic, Cleveland Clinic and Intermountain Healthcare, as well as the nation’s largest public sector systems, the U.S. Military Health System and the Veterans Health Administration – bear a special responsibility for addressing unwarranted variation. First, because these systems are organized, they are uniquely situated to adopt the principles of informed patient choice as a standard of practice, coordinate care for their chronically ill patients across sectors of care and, because they serve a defined population, control capacity and growth through population-based planning of resource allocation. Second, because they have the necessary infrastructure, including electronic medical records, they are uniquely situated to conduct the science of health care delivery: to develop and validate clinical pathways to rationalize the use of supply-sensitive care and assure high quality shared decision making. From such research, they (and the nation) will learn the resources (the dollars, the per capita workforce and facilities) needed to provide cost effective care for those with chronic illness and understand the demand for discretionary surgery and other forms of preference-sensitive care. Finally, organized systems of care, particularly those in the private sector, should accept the responsibility to grow: to enter into new markets and help the nation convert disorganized systems of care. In Chapter 13, I discussed some ideas as to how shared savings programs could create the incentives to make this happen.

The challenge to payers emerges from the special requirements for meeting the financial needs of providers who implement shared decision making, reduce overuse, and coordinate care for patients with chronic illness over time and across sectors of care – in other words, who are committed to reducing unwarranted variation. To meet these needs, the financial incentives affecting a given provider need to be coherent across payers: for example, policies governing the sharing of savings and other “pay for performance” strategies must be similar for all chronically ill patients and incentives to provide shared decision making must apply to all candidates for surgery, not just those who are covered by a particular payer. The challenge is thus to move away from the traditional volume-driven model for reimbursement and competition among payers based on discounts on the price per unit of service, to new models that support, indeed promote, providers who struggle to implement the principles of health care reform.

The Challenge to Patients and Families
Ultimately, the success of any reform effort to make informed patient choice the standard of practice, to reduce overuse, and to promote organized care will require support from a broad constituency. The natural appeal of shared decision making to the patient, combined with the efforts on the part of state legislatures to promote it, bodes well for accelerating the transition to informed patient choice as the standard of practice. Strengthening the role of the primary care physician as professional advocate for shared decision making and the introduction of federal legislation to support its implementation give me hope that one day, perhaps in the not-too-distant future, patients will no longer receive elective services that they would not have chosen had they understood the tradeoffs.

Gaining support for addressing the problems associated with the overuse of acute care in managing chronic illness will be the greater challenge. Providers who are threatened with the prospect that limits might be placed on their high-intensity patterns of practice might resort to what Robert Evans, the Canadian economist, calls “shroud waving,” raising the specter of health care rationing. Indeed, some already have. The epidemiologic evidence that greater care intensity is not producing longer life, higher quality of care, or greater patient satisfaction should help keep the debate focused; and Dartmouth Atlas data describing on a hospital-specific basis the patient’s experience of care at the end of life could help patients and families avoid hospitals that deliver unnecessary care.

But statistics, it has been said, “are people with the tears wiped off.” Getting beyond the “more is better” assumption will likely require a national debate on the limitations of medicine’s power to heal and cure, and on the quality of care at the end of life. The decisive voice in such a debate may come from the experiences of the Baby Boomers, who are about to enter the period of life when chronic illness begins to take hold. Many are already gaining firsthand experience with the problem of overuse as they struggle to help their aging parents cope with chronic disease and care at the end of life.

Reforming our health care delivery system requires a transition from today’s mostly disorganized care to organized, coordinated systems of care, and from delegated or “rational agent” decision making to shared decision making and informed patient choice. This will not be easy. After all, it requires transforming the culture of medicine and reengineering an industry that accounts for nearly 18% of the U.S. gross domestic product. But such is the eye of the needle through which we must pass to achieve significant reform.

Copyright © 2010 Oxford University Press. All rights reserved.

Tracking Medicine: A Researcher's Quest to Understand Health Care