US Health Care Costs – Why So High?


The US pays far more for health care than any other nation but gets less. On generally accepted indicators of health care quality, the US ranks at the bottom of the OECD country rankings[1].

Why is US health care so expensive and why are results so bad? In a recent article, I speculated costs were high because of:

  • back office paperwork resulting from so many insurance providers;
  • keeping old people alive too long in hospitals, and
  • doctors who are unwilling to delegate patients’ first meetings to lesser qualified health care personnel.

And clearly, the US obesity epidemic is driving up health costs.

But what do I know? I am an economist and not a health care expert. So I over the next few months, I will be interviewing a group of health care professionals for their takes on this question.

Jo Anne Magee has spent 35 years in the health sector. After 5 years working as a nurse for the United Mine Workers in West Virginia, Jo Anne got a Masters of Health Services Administration degree (MHSA) from the University of Michigan. Since then, she spent more than 25 years in insurance companies followed a stint with a care management startup company.

When I approached her on this question, she asked if I knew anything the “Practice Variations” research.  I said “No”.

An Economic Hypothesis

Before getting to “Practice Variations”, let me offer an economic hypothesis on US health care management. Most health care costs involve hospitals, doctors, and their equipment. What do hospital managers and doctors do? Try to provide patients with the best service at the lowest possible cost? No. I hypothesize that they focus on generating revenues to pay for its “products”. Knowing what health insurance will pay for, they focus on:

  • Filling hospital beds;
  • Getting patient appointments for its doctors;
  • Getting referrals for its specialists;
  • Getting billed time for its diagnostic test machines[2];
  • Scheduling imaging exams, and
  • Getting patients for its intensive care units.

A further observation on hospitals: what industries are the biggest real estate developers in cities? Universities and hospitals. Why? Because with their high cash flows, it is easy for them to get financing for more buildings.

To summarize, I hypothesize that hospital managers, working with doctors:

  • maximize hospital revenues with little regard for health care quality;
  • do anything insurance will pay for, and
  • become real estate developers.

Is there any evidence for this?

Practice Variations

John Wennberg and others have made “Practice Variations” research a promising vehicle for cutting health care costs. So time ago, Wennberg had a research grant that allowed him to collect data on how health care was delivered to Vermont towns. What Practice Variations did he find?

  • the rate of tonsillectomy was 60% and 20% in adjacent towns;
  • uterus surgery rates varied among towns by more than fourfold, gallbladder surgery by threefold, while
  • hospitalization rates varied by three times for respiratory disease and two times for digestive disease.

These findings lead to further work on practice variations. Many of these variations have been mapped for the country online in the Dartmouth Atlas of Health Care.

Wennberg and his co-workers did an in-depth study of how patients with a chronic illness were treated in their last two years of life[3]. When looking at US health costs, this sample is significant:

  • More than 90 million Americans live with at least one chronic illness, and seven out of ten Americans die from chronic disease.
  • Among the Medicare population, the toll is even greater: about nine out of ten deaths are associated with just nine chronic illnesses.
  • As chronic disease progresses, the amount of care delivered and the costs associated with this care increase dramatically. Patients with chronic illness in their last two years of life account for about 32% of total Medicare spending, with much of it going toward physician and hospital fees associated with repeated hospitalizations.

What were the findings? Remarkably little correlation between the prevalence of severe chronic illness and per capita Medicare spending across regions. The study demonstrated a nearly threefold variation across hospitals in dollar spending, average hospital days, and physician visits.

For example, at Duke University Hospital, considered one of the best in the country, chronically ill patients dying between 2001 and 2005 spent, on average, only 3.4 days in the ICU in their last six months of life. In contrast, patients at the UCLA Medical Center, also considered one of the nation’s best hospitals, spent, on average, more than 11 days in the ICU during the last six months of life.

Why would such variations occur? Wennberg claimed the “variation …cannot be explained by illness, medical evidence, or patient preference”. One might think it had to do with the severity of the disease. But Wennberg studied such a large sample that differences in severity would have averaged out. More expensive diseases? No. The 2008 Atlas compared costs across hospitals for patients with similar chronic illnesses.

How does he explain such variations? He argues that health treatment is “supply sensitive”. He quotes Milton Roemer, a health policy researcher: “A built hospital bed is a filled hospital bed”. Wennberg finds that:

  • the supply of hospital beds explains 54% admissions for all medical conditions;
  • the supply of cardiologists explains 49% of cardiac appointments.

In short, Wennberg questions whether patients who receive more supply-sensitive care[4] have better outcomes, live longer, or have a better quality of life. He points out that doctors and patients tend to believe that more care is better. He concludes[5]:

  • Supply-sensitive care accounts for well over half of Medicare spending, with most going to patients with severe chronic illness;
  • Whether from the patient’s perspective (satisfaction, technical quality, health outcomes) or from physicians’ perspective (quality of communication among physicians, continuity of care), higher spending and greater use of supply-sensitive care is not associated with better care;
  • Greater per capita use of supply-sensitive care and more spending do not result in lower mortality or improved quality of life; nor do they lead to improvement in the quality of care;
  • If the US modeled its care on efficient providers such as the Mayo Clinic, the Geisinger Clinic, and the Cleveland Clinic, we could shave 30% to 40% off the cost of caring for Medicare’s chronically ill patients;
  • Overtreatment harms patients, and it contributes to the chaotic quality of American health care. Also, overtreatment wastes taxpayer dollars. Various estimates for the amount we waste on overtreatment in this country range between 20 to 30 cents on every health care dollar spent.

To provide a better sense of the practice variations Wennberg found, I offer several tables. Table 1 provides the number of days spent in the hospital in the last two years of life for the 5 states with the highest and lowest rates. It is hard to believe that the 15 day difference in hospital stays between Utah and New Jersey is warranted.

Table 1. – Hospital Days, Last 2 Years of Life, by State

State Days
Utah 11.6
Oregon 12.0
Idaho 12.3
Washington 12.9
Montana 14.0
Mississippi 23.7
Hawaii 25.0
District of Columbia 26.0
New York 27.1
New Jersey 27.1


But let us go further and consider the differences between the hospitals. Table 2 provides the number of days spent in the hospital in the last two years of life for the 5 hospitals with the highest and lowest rates. It is certainly hard to understand the 49 day differential between Brigham City and South Beach.

Table 2. – Hospital Days, Last 2 Years of Life, by Hospital

State Hospital Days
Utah Brigham City Community Hospital 8.6
Idaho Walter Knox Memorial Hospital 9.0
Oregon Cottage Grove Community Hospital 9.6
Oregon Pioneer Memorial Hospital 9.7
Oregon West Valley Hospital 9.7
New York Cabrini Medical Center 54.5
New York Kingsbrook Jewish Medical Center 55.0
California Temple Community Hospital 55.6
New York St. John’s Episcopal Hospital 56.7
Florida South Beach Community Hospital 57.7


Consider now the second data set that Wennberg used in constructing his health care intensity index – the number of different physicians in the last six months of life. Table 3 provides the data for the 5 states with the highest and lowest numbers. So in Montana, patients who were to die in 6 months saw 6 new doctors while in New Jersey, they saw 10 more doctors. And this is after controlling for different chronic diseases.

Table 3. – Physicians’ Visits, Last 6 Months, by State

State Visits
Montana 4.96
Wyoming 5.08
Alaska 5.15
Idaho 5.23
Utah 5.32
Nevada 9.05
Pennsylvania 9.17
New York 9.24
Delaware 9.32
New Jersey 10.35


The data appearing in Table 3 are presented in Table 4 for new doctors by the hospitals with the highest and lowest numbers. Perhaps I am overly cynical, but it is hard for me to understand why patients in the Frankford Hospital have to see 16 new doctors in the last 6 months of life, unless it is a way to generate Medicare reimbursable income for the doctors.

Table 4. – Physicians’ Visits, Last 6 Months, by Hospital

State Hospital Visits
Oklahoma Integris Blackwell Regional Hospital 2.52
New York Bellevue Hospital Center 2.67
New York Kings County Hospital Center 3.14
Illinois John H. Stroger Jr. Hospital 3.47
South Dakota Huron Regional Medical Center 3.57
Florida Delray Medical Center 15.30
Pennsylvania Jeanes Hospital 15.31
New Jersey Good Samaritan Hospital 15.45
Pennsylvania Nazareth Hospital 16.35
Pennsylvania Frankford Hospital 16.81


Wennberg’s findings are consistent with the hypothesis I posed at the outset – that hospital managers and doctors work to generate as much income as they can without regard to what patients really need.

As would be expected, Wennberg’s work done with partners at Dartmouth has generated considerable debate and opposition[6]. Much of the opposition stems from not understanding the methodologies employed. But some comments are notable. For example, a more limited study found that patients were kept in hospitals far longer than was needed awaiting post discharge facilities (primarily nursing homes)[7].

As an economist, I wonder about possible economies of scale. Perhaps limited clusters of medical expertise might accelerate the growth of knowledge even though patients would have to travel further to benefit. Patients being treated in such “clusters” should expect to be incorporated into the on-going research of participating medical schools. And one would expect more doctors’ visits.


One can quibble about various aspects of Wennberg’s methodology. But his conclusion is unassailable: there are tremendous inefficiencies in the US health care delivery system resulting from supply-driven care. Wennberg estimates that these inefficiencies constitute 30% of end of life health care costs.

So my guesses for inefficiencies in the US health care system now include:

  • back office paperwork resulting from so many insurance providers;
  • keeping old people alive too long in hospitals;
  • doctors unwilling to delegate patients’ first meetings to lesser qualified health care personnel;
  • the US overweight/obesity epidemic, and
  • supply-driven care.

And I wonder about the importance of family support, particularly in urban areas. Unlike other countries where children assume responsibility for their parents in later years, most old people “fend for themselves” in the US.

And how about malpractice fears and how they influence hospital and doctor decisions?

So far, very little has been said about whether these inefficiencies can be rectified and if so, how. This topic will be the focus of the next article in this series.

And oh yes, we should not forget the role of the uninformed patient. A retired doctor friend told me the following story:

Some years ago I participated in a study of otitis media (swelling of middle ear) where it was found that only one third of the patients had the bacterial form (and hence could benefit from antibiotics).

It was further found that those given antibiotic treatment returned to normal in about ten days. It was also found that those not treated returned to normal in about ten days.

Conclusion: There is no way to explain this to Mother. Give her a prescription for her child’s antibiotic.

[1] T.R. Reid’s book on health care in other countries – “The Healing of America: A Global Quest for Better, Cheaper, and Fairer Health Care” – demonstrates that cheaper and better health care is possible.

[2] It is troubling to note that testing kills: Recent research shows that radiation from CT scans will eventually kill thousands of patients a year.


[4] Supply sensitive: more beds – more hospital patients, more doctors – more doctors appointments, more MRI machines – more MRI scans, etc.

[5] John Wennberg, “Tracking Medicine: A Researcher’s Quest to Understand Health Care”, Oxford University Press, 2010.

[6] For a criticism of the Wennberg methodology, see Peter B. Bach, “Resurrecting Treatment Histories of Dead Patients, A Study Design That Should Be Laid To Rest,” JAMA, 2004; 292: 2765-2770. For a debate on the issues, see the discussion in the “New York Times”:

[7] Gerald W. Neuberg, “The Cost of End-of-life Care, A New Efficiency Measure Falls Short of AHA/ACC Standards,” Circulation: Cardiovascular Quality and Outcomes, 2009; 2: 127-133.

The content above was saved on the old Morss Global Finance website, just in case anyone was looking for it (with the help of
This entry was posted in Global Economics, Global Finance. Bookmark the permalink.

Leave a Reply

Your email address will not be published. Required fields are marked *