Off The Charts
Hospital admissions and lengths of stay vary widely in Vermont, and a new report asks why
You’ve got a nasty infection that won’t heal, so you go to your local ER. The doc examines you, shoots you up with antibiotics, and sends you home with instructions to visit your primary physician for follow-up. Then you talk to your friend who lives a couple of hours to the south. Something like that happened to him last month, he says — only his hospital admitted him for overnight observation. They ran a lot more tests and had him see a specialist, too.
Everyone can cite anecdotal differences between this hospital and that hospital. Question is, what’s behind those differences? And, more important, which patient is ultimately getting better care, the one who’s fussed over or the one who’s sent home?
The second question’s a toughie. But an answer to the first one comes from a new “white paper” that examines recent data on the rates at which Vermonters are hospitalized and their average lengths of stay. The data reveal that in health care, as in retail, it’s all about location. In Vermont, the likelihood that you’ll be admitted to a hospital, the procedures you’ll get and how long you’ll stay often have as much to do with where you live as with your health.
The statistics, culled from the Vermont Department of Health and the Department of Banking, Insurance, Securities and Health Care Administration, reveal “significant variations from region to region” at Vermont’s 14 community hospitals. Moreover, these variations appear to stay consistent even when the numbers are adjusted for differences in the size, average age and health status of the patient population. Adjusting the data for age, a key indicator in how sick a patient is likely to be, is particularly important because it directly contradicts the explanations some hospital administrators offer for the stats of their institutions.
The white paper — titled “High Use in Vermont Hospitals: It’s Costly, But Is It All Necessary?” — was released this week by Burlington-based health-care consultant Jeanne Keller, of Keller and Fuller, Inc. Examining data from 2005, the last year for which figures are available, Keller found that residents of the Rutland “hospital service area” (HSA) were admitted at a rate of 109.3 hospitalizations for every 1000 residents — considerably higher than the statewide average of 85.7 admissions per 1000 residents. Residents of the Morrisville HSA were admitted at the lowest rate in Vermont — 73.9 per 1000 residents. Burlington-area residents were hospitalized far less frequently than the statewide average, at a rate of 77.6.
Keller also found that people who lived near hospitals with higher rates of admission tended to spend more time in those hospitals. For instance, patients in the Rutland area collectively spent about 538 days in the hospital for every 1000 residents; those in Springfield spent 448 days; Bennington, about 442 days. These average lengths of stay were considerably longer than the statewide average of 398 days. Vermonters served by the two largest hospitals in the region — Fletcher Allen Health Care in Burlington and Dartmouth-Hitchcock Medical Center in Lebanon, N.H. — spent far less time in the hospital than the statewide average — 379 and 355 days, respectively.
Keller’s white paper asks more questions than it answers. But she says that’s the point: to lay the groundwork for more detailed investigations and analyses. Regardless of the causes, these findings have serious ramifications as lawmakers look for new ways to expand health-care coverage to more and more Vermonters. And these numbers are particularly relevant now that Vermont’s per-capita medical costs are rising at the highest rate of any state in the country, according to the latest figures from the federal Centers for Medicare and Medicaid Services.
Keller wants to pose some hard questions about what’s causing these regional differences and what they’re costing Vermonters in terms of patient safety, decreased productivity and diminished quality of life. Going back to our anecdote: Is the higher rate of hospitalization in your friend’s community a sign of underlying public-health problems there, or an indicator of an “overly aggressive” style of medical care? Conversely, if you live in a community with much lower rates of admission and shorter hospital stays, such as Morrisville and Randolph, are you not getting the level of medical care you need?
Keller theorizes that the introduction of new medical services at some hospitals — such as specialized-care units and state-of-the-art diagnostic and imaging equipment — could be creating pressure to admit more patients to those hospitals and keep them there longer. She concludes that people in high-use areas may be subjected to “unnecessary” and “dangerous” medical and surgical procedures, including the added risks of infections and medical errors.
Considerable research done by the Center for Health Policy Research at Dartmouth Medical School suggests that patients who receive more aggressive care don’t always fare better. In fact, quite often they fare much worse. As one Dartmouth-Hitchcock researcher puts it, “Hospitals can be dangerous places, especially if you don’t need to be there.”
Ultimately, Keller suggests that “high-use” regions should bear the ultimate responsibility for (read: cost of) these variations in hospital care, especially considering their financial impact on public and privately funded insurance plans. According to her calculations, if patients in Vermont’s three “high-use areas” — i.e., Rutland, Springfield and Bennington — had been hospitalized at a rate comparable to the statewide average in 2005, the savings would have totaled $20.6 million.
“The horrible problem we’ve got,” Keller says, “is that nobody in the state is given the job, the responsibility or is held accountable for asking these questions — and then getting them answered.” She suggests creating an independent body, similar to the “think tanks” in other states, that would routinely examine these numbers on behalf of the public, then challenge hospitals that may be providing “inappropriate” care.
But not so fast, warn several spokespeople for the hospitals and the state’s hospital association. They contend that these figures, though beneficial for certain statewide comparisons, don’t necessarily tell the whole story. They argue that it’s not always useful, or even fair, to compare one Vermont hospital to another, since they differ so much from one another in size, budgets, services offered and populations served.
Kevin Robinson is communications director for Southwestern Vermont Medical Center (SVMC) in Bennington, a hospital service area that ranks near the top for admission rates and average lengths of patient stays. Robinson, who spent a lot of time with Seven Days trying to tease out what his region’s high numbers mean, suggests underlying demographic factors should be considered as well.
“In some respects, it’s apples and oranges,” Robinson says. “I’m not sure if it’s fair to compare a Fletcher Allen to a Copley [Hospital].” Copley, a 63-bed community hospital in Morrisville, had the lowest rates of admission and the shortest patient stays of any hospital in the state. In contrast, Fletcher Allen is a 500-bed, tertiary-care medical center. It has the third lowest rate of admissions in the state.
To account for the high percentage of patients admitted to SVMC through its emergency department, Robinson points to the relatively large number of nursing home beds in the Bennington area, including the Vermont Veterans Home. According to the white paper, Bennington had the highest rate in Vermont — 68 percent — of patients hospitalized through its ED. That’s well above the statewide average of 48 percent in 2005.
The availability of good hospice care is one factor that can affect patient stays: Robinson suggests that, with relatively few such options, Bennington-area residents may seek end-of-life care in the hospital rather than at home. Income level is another piece of the puzzle. While poorer communities tend to have more ailments than wealthier ones, Robinson says low-income people often lack the means to get to the hospital, which may keep their admission rates down.
“At the end of the day, these numbers probably tell you as much about the populace as they do about a particular hospital,” Robinson says. “Admissions to the hospital are based on clinical criteria. Someone shows up at the hospital and they’re sick, and the doctor assessing them decides whether they need to be hospitalized.” SVMC has clinical case managers who review all admissions. Robinson says they’ve seen “no indication” that Southwestern is admitting patients who don’t need to be there.
Other hospitals offer similar explanations. Larry Jensen is vice president for corporate development at Rutland Regional Medical Center. As he puts it, “Generally speaking, hospitals don’t admit patients. Doctors do.”
Jensen says he’s seen the numbers Keller based her report on, and agrees they are “statistically significant.” However, like Robinson, he suggests the numbers say more about the doctors and patients in the Rutland community than about the hospital itself. As he points out, more than 60 percent of Rutland Regional’s patients are on Medicare or Medicaid, which could speak to the level of primary care they receive.
Jensen confirms that Rutland Regional has been the site of significant capital projects and growth in recent years, including a new CT scan, an $8 million emergency department overhaul and a newly proposed $25 million oncology unit. However, he dismisses Keller’s suggestion that Rutland’s high rates of hospitalizations and longer patient stays are a sign of “unwarranted admissions” or “unnecessary procedures.”
“We have conducted a number of improvement projects around lowering the length of [patient] stays,” Jensen says, “and we continue to follow our lengths of stay very closely.” He adds that these projects were implemented only in the last two years — so any declines in Rutland’s rates wouldn’t be reflected in 2005 figures.
Bea Grause is president and CEO of the Vermont Association of Hospitals and Health Systems. She has also seen the state figures Keller quotes. But she agrees with Jensen and Robinson that there are other possible explanations for these regional variations. They could include the mix of professionals in the hospital, differing physician practices, the number of discharge coordinators and nurse staffing.
“Ultimately, it’s the physician who decides to admit,” Grause says. “They don’t seek approval from a hospital CEO or administrator.”
But Keller counters that, quite the contrary, hospital boards and administrators are the ones who set the standards of care for their own facilities, and they decide which physicians are granted admitting privileges. By and large, she says, Vermont’s hospitals really only compete with one another “in the marketplace of public opinion.” As a result, if some doctors and emergency departments are more inclined than others to fill empty beds and perform costly and potentially unnecessary tests and procedures, only “benchmark humiliation” will make them re-examine their medical practices and reform their ways.
Keller is a well-known health-care policy analyst, and her work over the years has led to a number of significant policy reforms on behalf of Vermont consumers. She’s also long been a thorn in the side of the state’s hospital industry, and her white paper will undoubtedly be greeted in some circles with a degree of skepticism.
In 2003, for instance, Keller was instrumental in the passage of Act 53, which forced hospitals to publish annual “community report cards” on patient safety, infection rates, nurse staffing levels and other safety and financial measures. Eight years ago, Keller worked with Rep. Anne Donahue (R-Northfield) to help uncover the financial irregularities and cost overruns on Fletcher Allen’s “Renaissance Project,” which cost Vermont taxpayers tens of millions of dollars.
More recently, Keller pushed lawmakers to include an “adverse event reporting” provision in last year’s Catamount Health legislation. Under that provision, which took effect January 1, hospitals are now required to notify the state health department within 48 hours of certain preventable medical errors — such as leaving a sponge inside a surgical patient, performing an operation on the wrong patient or limb, or sending a newborn infant home with the wrong parents.
As it happens, Keller’s findings in this white paper are neither new nor groundbreaking. Medical researchers have known for decades that, as one doctor put it, “geography is destiny.” In other words, where a patient lives is frequently a determining factor in both the quality and quantity of care he or she receives. And the health-care industry owes this discovery — now an entire field of medical research known as “small-area variation analysis” — to a former Vermont physician who figured it out more than 30 years ago simply by looking at children’s tonsils.
In the early 1970s, Dr. John “Jack” Wennberg was director of the state’s Regional Medical Program at the University of Vermont. Wennberg lived halfway between Stowe and Waterbury, where he had children attending public school. At the time, only about 20 percent of the children in that school district had their tonsils removed by the time they were 15.
Curiously, Wennberg discovered that the situation was dramatically different just down the road in Stowe, where about 70 percent of all schoolchildren had their tonsils out by that age. Since there were no significant environmental, demographic or sociological factors to account for these geographic differences, Wennberg eventually reached the conclusion that the higher rates of tonsillectomies in some communities were driven by, in his words, “supplier-induced demand.” In short, it was the prevalence of certain types of medical specialists — namely, surgeons — in those communities that led to more frequent tonsil snipping.
But Wennberg didn’t end his research at children’s throats. He later uncovered significant regional variations for other medical and surgical procedures, including appendectomies, gall bladder removals, hysterectomies and Cesarean sections. For instance, he and fellow researcher Alan Gittelsohn discovered that in certain towns in Maine, men were four times more likely to have their prostate removed by age 80 than men in other towns. Likewise, doctors in New Haven, Connecticut, were twice as likely to perform heart bypass surgeries as those in Boston.
It was just a matter of time before the media picked up on Wennberg’s research. In 1984, WCAX-TV in Burlington ran a 25-part series called “Patterns of Practice,” which explored the wide disparities among Vermont hospitals in length of patient stays and frequency of certain surgical procedures. When it was revealed that hysterectomies were far more common in St. Albans than elsewhere in Vermont, reporter Hamilton Davis quipped that there probably wasn’t a 52-year-old uterus left in Franklin County. “Patterns of Practice” earned the station a Peabody Award and spurred a statewide investigation. Its findings eventually led to a 20 percent decline in average patient stays at hospitals throughout the Green Mountain State.
But 24 years later, not much has changed, Keller asserts. While the state continues to compile and publish these statistics — at significant cost to taxpayers — no independent body or state agency is charged with the task of analyzing them and then doing something about them.
“When is someone going to pay attention?” Keller asks. “This is how we can get a handle on health-care costs. There is still so much low-hanging fruit to pick before we start denying people necessary care.”
Dian Kahn, director of analysis and data management with the Vermont Division of Health Care Administration, agrees. As the person responsible for compiling these figures each year, she admits that it’s beyond the scope of her job to analyze what these numbers actually mean. “I think [these data] are a good way to show that there’s something here we need to look at more closely,” Kahn says. “It’s a way to bring good research and policy questions to the surface.”
Equally important, Vermont could use these findings to avoid millions of dollars in potentially unnecessary, and possibly even dangerous, medical procedures and hospital stays.
Dr. Elliott Fisher is director of the research center and the principal investigator on the Dartmouth Atlas Project — a program co-founded by Jack Wennberg that conducts comparative analyses among the nation’s 4700 or so hospitals.
Seven Days spoke to Fisher last week about the data Keller uncovered and asked him whether he thinks Vermont’s different rates of admission and average patient stays are driven by health factors within each community, or by forces at the hospitals themselves. Since Fisher had not seen Keller’s report, he couldn’t comment specifically on her numbers. And, based on the research he’s done in other parts of the country, he says there probably are some slight regional differences in health status from one community to another.
That said, Fisher doesn’t believe such differences account for the large variations in hospitalization rates and lengths of patient stays in the Vermont data. In fact, his research has found “pretty good evidence” that the number of hospital beds in a given area, relative to the size of the population, is often a “powerful influence” on how much that hospital is used. In other words, if you build the beds, patients will come.
In one study, for example, Fisher compared hospital usage in Sacramento, California — a city with few hospital beds relative to its size — to Los Angeles, an area with many hospitals and physicians. He found that the medical resources in Los Angeles were being used at a 60 percent higher rate than those in Sacramento. The potential savings to taxpayers over the five-year period of his study could have been “enormous:” Medicare would have saved $1.7 billion in L.A. alone had its pattern of care mirrored that in Sacramento.
Moreover, Fisher’s research with Wennberg has repeatedly shown that more aggressive care doesn’t lead to better patient outcomes. In another study, the researchers contrasted Medicare spending on chronically ill patients in New York City and Miami with spending on similar patients in Salt Lake City and Rochester, Minnesota. Researchers found that patients in NYC and Miami who received more aggressive (and expensive) care fared no better than those in Salt Lake City and Rochester. Often, their outcomes were much worse.
Fisher blames the current system of hospital reimbursement, which acts as a disincentive for hospitals to keep their admission rates down and focus on patients’ overall health. “We know that hospitals cannot afford to keep beds staffed and empty,” he says. “Right now, hospitals are punished if they keep people out of the hospital.” And, under Vermont’s current system of hospital budgeting, “If you don’t meet your revenue targets, you’re not going to make your bond payments.”
For her part, Keller sees these numbers as a useful tool for lawmakers, especially as they consider a bill — H.304 — that would create a statewide hospital insurance plan and cap hospitals’ annual rate of growth. Keller argues that the bill, though well intentioned, does little to address the underlying forces that are driving up health-care costs. In 1991, Vermont’s per capita health-care spending was 88 percent of the national average. In 2004, it was 115 percent. Neither the age of Vermont’s population nor the health status of Vermonters can explain that rapid rate of growth.
“Just like we need someone to make sure that our water is clean every day, we need to make sure that we’re not being hospitalized unnecessarily,” Keller concludes. “And if it’s not the government, it’s going to be the insurance companies. Which one do you think the people will prefer?”