Statement of John E. Wennberg, M.D., M.P.H., Director,
Center for the Evaluative Clinical Sciences, Dartmouth College, Hanover, New Hampshire

Testimony Before the Subcommittee on Health
of the House Committee on Ways and Means

Hearing on Promoting Disease Management in Medicare

April 16, 2002

My name is John Wennberg. I am a member of the faculty of Dartmouth Medical School. I have been asked to comment on variations in Medicare spending and quality, and on the implications of these variations for the design of chronic disease management programs.  It is now well known that spending for traditional, fee-for-service, Medicare varies extensively.  For example, in 1996, age, sex and race adjusted spending for non-HMO Medicare services in the Miami hospital referral region was nearly two and a half times higher than in the Minneapolis hospital referral region.  In a recent Health Affairs article (attached), my colleagues and I described some of the variations in quality of care among hospital referral regions, as well as the association between quality and spending.  We looked at three categories of medical services: “effective care,” “preference-sensitive care” and “supply-sensitive care.”

It is important to make distinctions among effective care, preference-sensitive care and supply-sensitive care because the causes of the variation and the remedies are different for each category.  Moreover, variation in per capita Medicare spending is associated with only one of these categories — supply-sensitive care.  If disease management programs are to have an important impact on the overall quality and cost of care for the chronically ill, the programs will have to address variations in each category. 

I will summarize very briefly what we mean by the terms effective, preference-sensitive, and supply-sensitive care.  Each of these categories of care raises a different challenge to the design and implementation of disease management programs.

Effective care is evidence-based care – care that we know works and we know all eligible patients should get.  Our studies show extensive underuse of effective care among fee-for-service Medicare enrollees in virtually every part of the country.  For example, in some regions of the United States, less than 20% of post-heart attack patients who were classified as “ideal” candidates for a particular medication actually received the medication.  There are similar patterns of underuse of effective treatments for Medicare enrollees with diabetes, and evidence that there is poor compliance with guidelines for cancer screening and immunizations.  The principal cause of variation in the delivery of effective care is a lack of infrastructure – systems that make sure that appropriate care is provided in a timely way.

Correcting the underuse of effective care will demand improvement in the infrastructure of the everyday practice of medicine.  Staff model HMOs provide exemplary models of how this can be done and a number of network HMOs (and their disease management company subcontractors) have had success in promoting provider compliance.  Health insurance claims provide the information databases that can identify patients who need services, so that both patients and providers can be reminded to seek and administer appropriate care in a timely way.  The processes of care can be monitored with these claims databases, at the level of individual practitioners or physician groups.  We have shown that this monitoring is possible using the Medicare claims data; and many health systems already have the information available, but have not put it to use.

Underuse of effective care can be reduced by collecting and monitoring the processes of care at the hospital and physician group levels.  Medicare claims data can serve as the basis for establishing disease management registries to identify patients in need and document health care quality.  Medicare claims should be made available in “real time” for use in disease management demonstration projects in traditional Medicare.

Preference-sensitive care pertains to conditions where at least two valid treatment strategies are available, each with its own risks and benefits.  Since the choice of treatment involves tradeoffs that ought to depend on patients’ preferences — such as the choice between surgical or pharmaceutical treatment — the patient’s choice of treatment should determine what is done.  There are wide variations in the use of discretionary surgery and other preference-sensitive treatments.  For example, the rates of lumpectomy for breast cancer and prostatectomy for prostate cancer vary by a factor of five among hospital referral regions in the United States.  In some cases, the principal cause of variation is that there is substantial uncertainty about the risks and benefits of the treatment options because there have been no clinical trials to determine the facts.  The fundamental problem, however, is the failure to involve the patient in a meaningful way in the choice of treatment.  Decisions should belong to the patient, but provider opinion tends to dominate.  For this reason, we believe there is widespread misuse of preference- sensitive care.

Dealing with the misuse of preference-sensitive care requires that patients become actively involved in the choice of treatment in those clinical situations where options exist.  Shared decision making or “informed patient choice” is gaining recognition as the remedy for unwarranted variations in this category, but implementation is just beginning.  Staff model HMOs and a few IPA HMOs provide examples of implementation, but the results are spotty.  A basic problem facing IPA and traditional Medicare is that the fee structure doesn’t reward providers who implement shared decision making; in other words, reimbursements reward activity, rather than careful consideration of all the options.  The most important goal of this effort should be to make sure that people with chronic illnesses know what the options are (including costs, risks, and benefits).  The current system does not reward or encourage physicians to engage in this effort.  It should, because it is the right thing to do.  It could also save money.  Several studies have shown that informed patients tend to want less surgery than surgeons are inclined to recommend under the current system of rewarding procedures rather than counseling.

Misuse of discretionary surgery and other forms of preference-sensitive care can be reduced by actively involving patients in the choice of care.  We must make sure that chronically ill patients understand their options and are aware of the benefits and risks of treatment.  Disease management demonstration projects should be given flexibility to deal with the current flaws in the traditional Medicare fee structure that perversely interfere with informed patient choice.

Supply-sensitive services are those in which the frequency of use is not determined by well-articulated medical theory, much less by scientific evidence.  Supply-sensitive services include physician visits, diagnostic tests, hospitalizations, and admissions to intensive care among patients with chronic illnesses.  The variations are particularly striking among the most seriously ill Medicare enrollees.  In some regions, Medicare enrollees average more than 20 visits to medical specialists during the last six months of their lives; in other regions, the average number of such visits is fewer than three.  In some regions, nearly 30% of Medicare deaths occur in ICUs; in other regions, fewer than 7% do.

The principal cause of variation in utilization of services is the regional and local variation in resources.  More hospital beds per capita mean more hospitalizations per capita among those who have chronic diseases such as congestive heart failure, diabetes and chronic pulmonary disease.  More cardiologists per capita mean more visits with cardiologists per person with heart disease.  Yet there is little medical theory — and no evidence — about the appropriate frequency of such services — that is, what rate of physician visits, for example, results in the best health outcomes.  Moreover, studies conducted at the population level have demonstrated that there are no gains in life expectancy associated with higher frequency of intervention.  For these reasons, we believe there is widespread overuse of supply-sensitive services.

Dealing with the overuse of supply-sensitive services poses the greatest challenge to the implementation of a comprehensive program in disease management.  Most of the care given to patients with chronic disease belongs in this category.  Those who live in areas with more resources are likely to have more visits with doctors, more tests, and more admissions to hospitals and to intensive care, particularly at the end of life. This is the category of care that “explains” higher per capita spending, but it is not associated with better quality of care or extension of life expectancy.  Medicare enrollees living in regions where per capita spending is higher than average don’t receive more effective care — and the differences in spending are also not explained by higher rates of major (non-elective) surgery.  (See Exhibit 2 in the attached article.)

The role of supply-sensitive services in both spending and outcomes has received little attention from academic researchers.  The topic barely makes it onto the research agenda, and there are few studies that could provide a basis for constructing guidelines that might tell us how many physicians are actually needed in a given region, what rate of visits to physicians results in the best outcomes for those with chronic diseases, or whether the use of expensive diagnostic technology actually improves accurate diagnosis, health, or survival.  Recent research on the redesign of clinical practice to introduce group visits and open appointment schedules is encouraging because it breaks the cycle of supplier-induced demand and could serve as the basis for redesign of chronic disease management. 

Except for staff-model HMOs, insurers and providers have not responded to the challenge of determining what works and how much is enough.  Much of the intellectual leadership in the research on redesign of clinical practice comes from these organizations, which are able to control their own per capita resources (eg. doctors hired, hospital beds constructed or contracted for, patterns of practice).  But the efficiency of staff-model HMOs in the use of supply-sensitive care is probably more a byproduct of a corporate strategy of remaining competitive with fee-for-service medicine than a conscious attempt to manage chronic disease.  As Alain Enthoven has pointed out, such HMOs practice “private sector health planning.”  That is, they regulate the capacity of their organizations.  Since the staff-model HMO standards for resources are consistently lower than the prevailing rates of resources in the regions where they compete, the cost of HMO care has been correspondingly lower than the cost of fee-for-service medicine.  For the nation to achieve HMO-like reductions in spending, health care organizations would need to become accountable for capacity.  This is a formidable task and one that in the absence of reform of payment systems is probably impossible.

Addressing the overuse of supply-sensitive care is made even more challenging by the cultural assumption that more health care is better.  There might be hope in the growing public awareness that for many patients with advanced chronic illness, high-technology interventions such as admissions to intensive care are both futile and unpleasant, degrading the quality of life in its closing months and years.  The chronic disease management movement might find natural allies among the growing numbers of Americans who are trying to reform of end of life care.

Reducing the overuse of hospitals, intensive care, physician visits and testing among chronically ill Americans should be a national priority.  This must be done if we want to reduce unwarranted Medicare spending and improve heath care quality.  Disease management programs must address this priority.  However, the scientific basis for defining optimal care is weak and, with the exception of staff model HMOs, the tools available for dealing with the causes of variation are also weak.  Demonstration projects are needed to establish benchmarks for effective clinical practice for supply-sensitive care and to identify “best practice” models for efficient care. 

While the task of comprehensive reform of the management of chronic disease is a formidable challenge, we have a moral obligation to take it on.  I applaud the committee’s willingness to address the complexity of the task and confront the difficulty of finding workable solutions.  Progress often begins with debate, which I hope these hearings will spark.  I also applaud the efforts by CMS to invite chronic disease management companies to implement experimental chronic disease management in fee-for-service Medicare.  We need to learn how well the IPA model can improve the quality of care in each category of service.  The complexity of the task, however, warrants additional demonstration projects, and some should be carried out by provider organizations serving traditional Medicare populations.  In our Health Affairs article, my colleagues and I recommend a demonstration project that could prove that the federal government and responsible health care organizations can establish a partnership to fruitfully address each category of unwarranted variation.  The results could, I believe, improve the quality and efficiency of services provided for all fee-for-service Medicare beneficiaries.  And it is only the first step in improving the quality of care for all chronically ill Americans while we demonstrate that doing medicine right does not mean doing it the most expensive way.