Hospital-stay guidelines: Just plain weird

August 6, 2001

Milliman's USA's advisories on hospital stays are coming under intense attack. How insurers use them may be as problematic as their scientific validity.


Cover Story

Hospital-stay guidelines: Just plain weird

Jump to:Choose article section...Milliman says guidelines describe targets, not norms It's hard to follow the guidelines in a sickly health system Do health plans apply the guidelines as intended? An odd fish in the guideline pond "These guidelines wouldn't hold up in court" Betting on managed care—and getting burned


Milliman USA's advisories on hospital stays are coming under intense attack. How insurers use them may be as problematic as their scientific validity.

By Robert Lowes
Midwest Editor

Back in the heyday of HMOs, health care actuary David Axene gave speeches saying that up to 60 percent of hospital care could be avoided. Either patients could be treated elsewhere, like at home, or they didn't need any care at all.

Someday, predicted Axene, perhaps the only patients that hospitals could count on would be those requiring intensive care or cardiac care.

"All others are at risk," he said.

Axene's old company, Seattle-based Milliman USA, has been leading this movement to empty out America's hospitals. An actuarial firm, Milliman publishes proprietary health care guidelines that it licenses to insurers covering close to 100 million Americans. These guidelines are best known for specifying how many days a patient with a particular illness should spend in the hospital. To insurers, the guidelines represent a powerful cost-control tool. Many doctors, however, see them as a flimsy excuse to discharge patients prematurely.

Let's say you admit a patient in cardiogenic shock to the ICU. He's diagnosed with congestive heart failure. In a best-case scenario, Milliman says, you should be able to discharge him that day or the next. Keep him in four or five days longer, as most doctors do, and you might get a phone call from a health plan's case manager asking, "Why is Mr. Jones still in the hospital?"

Such guidelines, frequently derided as unrealistic, have helped foment the backlash against managed care. In 1996, for example, the protest over Milliman's guideline of 24 hours in the hospital for mothers and newborns following normal vaginal delivery helped spur Congress to mandate a minimum two-day stay.

Insurers once bore the brunt of the backlash, but these days Milliman (until recently known as Milliman & Robertson) is getting hit between the eyes. A class-action suit against insurers such as Humana and Aetna USHealthcare in a Miami federal court accuses them of using guidelines from Milliman and other companies to defraud patients and doctors.* In another class-action suit, against Prudential Insurance Company of America, New York patients claim the insurer wrongly relied on Milliman guidelines to deny them needed care (see "Managed care on trial"). And four Texas pediatricians are suing Milliman for listing them erroneously as contributors to Milliman pediatric guidelines that they call dangerous (see below).

Faced with bad publicity, financial losses, and managed care's retreat from tight utilization management, Milliman's guideline business appears to be even more endangered than the hospital beds that David Axene described. Not surprisingly, Milliman, whose executives declined to be interviewed for this story, has been trying to sell its guideline division.

To be sure, some doctors praise the guidelines as a way to ration health care resources that too often are squandered by physicians, hospitals, and patients. They say Milliman is unjustly taking the heat; that the problem lies not with the guidelines, but with rigid insurers who treat ideal hospital stays as the norm. "I can't think of a more competent company," says internist John Santa, administrator of the Office for Oregon Health Plan Policy and Research. "Their product is well-researched, but like anything else, the user has some responsibility. "

Milliman says guidelines describe targets, not norms

Whether it's home care, pediatrics, or workers' compensation cases, Milliman has a guideline. It publishes eight different sets, the oldest of which covers inpatient and surgical care. Here you'll find the length-of-stay goals. But these guidelines do more than count days. They set criteria for hospitalizing patients in the first place. And when admission is advised, they chart the day-by-day course of an uncomplicated patient who's making the best recovery possible.

Again, consider the patient with congestive heart failure. On Day One, he receives diagnostic tests, among them an ECG and a chest X-ray. Therapies such as parenteral medications and oxygen commence, as does discharge planning. On Day Two, the patient is supposedly alert and meets key indicators of clinical stability, such as the resolution of pulmonary edema. Medications are switched from parenteral to oral delivery. The doctor may order more tests, after which the patient leaves the hospital—in theory, at least.

Milliman guidelines come with warning labels, though. While they apply to most patients in most situations, they can't be imposed automatically on any individual patient. The physician has the last word. "Use of guidelines requires, and never replaces, clinical judgment," Milliman says. And a Milliman day is not necessarily 24 hours. If that CHF patient still has pulmonary edema two days after he was admitted, he's still stuck on Day One of the guideline.

The recoveries envisioned by the guidelines have struck some experts as too good to be true. While Milliman sets forth a one-day stay or less for CHF patients, the mean nationwide is 5.5 days, according to Solucient, a health care research company in Evanston, IL. Only 8.2 percent of such hospital cases meet the Milliman guideline.

Solucient's predecessor company, HCIA-Sachs, questioned the realism of Milliman's pediatric guidelines in two reports several years back. Among 45 pediatric conditions, 64 percent of uncomplicated cases had average actual lengths of stay that exceeded the Milliman goal. The length of stay for uncomplicated bacterial meningitis, for example, was 8.5 days, compared to Milliman's three days.

"Apples and oranges" is Milliman's customary reply to such studies. The Solucient/HCIA studies, they say, merely report actual performance. In contrast, Milliman guidelines set a target for performance based on observed best practices across the country, as well as medical research.

That's all very well, say Solucient and others, but they also pick another bone: Milliman bases its guidelines on an uncomplicated patient, but is that a meaningful model? "Roughly three-fourths of hospitalized patients are indeed uncomplicated," says David Foster, a clinical informatics expert at Solucient, "but that rule of thumb doesn't apply to every illness. With bacterial meningitis, the percentage decreases."

Milliman guidelines do a better job of factoring in complications than they used to. The most recent edition lists reasons why patients might need a longer stay than suggested. Still, that's not good enough for Jacksonville, FL, endocrinologist Yank Coble Jr., the AMA's president-elect. "Who at the insurance company decides whether a patient is complicated?" asks Coble. "It's usually someone not qualified to make the call."

It's hard to follow the guidelines in a sickly health system

Another key Milliman assumption is that a hospitalized patient, besides being uncomplicated, enjoys grade A health care. Some guidelines, for instance, recommend a greater reliance on home health care to reduce hospital stays. But if home health care isn't readily available in a given community, the target length of stay may be unreachable, says Milliman.

And what about hospital diagnostic services? If CT scans or GI tests are unavailable on weekends, a doctor may not be able to discharge a patient according to the guideline timetable.

"Using Milliman guidelines requires systems support," says Rockville, MD, pediatrician Dennis Kane, director of hospital management services for Kaiser's Mid-Atlantic Permanente Medical Group. Kane says Kaiser's been able to provide that support, by and large. "We've had a good experience with the guidelines. If a hospital or community doesn't have the necessary services, it's not the guidelines' fault."

Chicago internist Robert Parker knows probably better than anybody what it takes to implement Milliman recommendations. When he was CEO of the 300-physician Carle Clinic Association in Urbana, IL, an affiliated health plan and hospital relied on the guidelines. Parker says it was easy to meet them, largely because "we had a solid, efficient infrastructure—radiology, labs, everything." Now Parker is CEO of the physician arm of Sinai Health System in Chicago, where the resources aren't as robust.

"It takes longer to get X-rays done because sometimes we don't have enough transporters to take patients from the floors to radiology," he says. "That plays into our ability to get a patient out of the hospital in 24 or 48 hours."

Do health plans apply the guidelines as intended?

Taken at face value, Milliman guidelines paint a health care paradise. The patient's condition dictates the care he receives, the services he needs are available when he needs them, insurers prune unnecessary hospital days, and the doctor's clinical judgments are always respected.

But what about the health plans that pay Milliman to use the guidelines? Do they remember all the caveats, all the assumptions about uncomplicated cases and available services?

"Sometimes they do, sometimes they don't," says Robert Parker. "There's not much consistency in the insurance industry, or even in one company," he says. "Nurse reviewers come and go. How they interpret the guidelines depends on their knowledge and experience. And some of them can be real policemen."

One aspect of the guidelines that's frequently misunderstood is the notion of the Milliman hospital "day." Remember, a patient doesn't graduate from Day One to Day Two unless he meets all of Day One's criteria—even if he's been in the hospital for a week. However, "hospitals have repeatedly told us that health plan nurses interpret a day as strictly 24 hours," says Lillian Forgacs, associate vice president for utilization management and managed care at the Greater New York Hospital Association. "That's where we run into trouble."

Such misinterpretations aren't surprising, given that teaching employees to use the guidelines properly hasn't always been a high priority of insurers. "Some clients who purchased or licensed the guidelines were unwilling to spend more money on training," says Mill Valley, CA, orthopedist Robert Bruckman, a former Milliman employee who developed its workers' compensation and return-to-work criteria.

Then again, Bruckman—now an independent consultant—says some insurers may have ignored warnings on the guidelines and used them as a pretext to trim hospital costs. Adds another Milliman alumnus: "I heard over and over that managed care organizations wanting to reduce bed days used the numbers as hard-and-fast."

An odd fish in the guideline pond

While Milliman guidelines influence the care of millions of Americans, the company doesn't present them as clinical practice guidelines per se (like those issued by a medical society) or a standard of care. Rather, their intellectual father, internist and former Milliman employee Richard Doyle, has called them a "standard of efficiency."

In the realm of true clinical guidelines, Milliman is an odd fish. Consider the nonprofit National Committee for Quality Assurance in Washington, DC. The NCQA evaluates how managed care organizations apply clinical guidelines, but its definition of a guideline doesn't include what Milliman publishes, because Milliman is more focused on utilization management.

You also don't find Milliman guidelines listed with the National Guideline Clearinghouse, sponsored by the federal Agency for Healthcare Research and Quality in partnership with the American Association of Health Plans, and the American Medical Association. "None of them would be acceptable," says the AMA's Yank Coble, who sits on the clearinghouse board. "They sound more economically oriented than quality-oriented."

The clearinghouse has never formally rejected Milliman's guidelines, because the company has never submitted them for inclusion. On the surface, there's nothing in the clearinghouse's guideline criteria that would automatically exclude Milliman. And it's not as if the clearinghouse bars the door to proprietary guidelines, says director Jean Slutsky, because organizations represented in it include a few private companies along with groups like the American College of Cardiology. However, to be included, private companies must allow the clearinghouse to publish detailed summaries of their guidelines—something close-mouthed Milliman is unlikely to do. After all, you can't even find its guidelines in a medical library. To see them, you have to rent them.

Coble's biggest beef with Milliman guidelines is that they're not clearly evidence-based. Milliman would beg to differ—the words "evidence-based" appear all over the company's literature. However, they haven't always made that claim.

Originally, the guidelines arose from consensus—Richard Doyle surveying other doctors in the 1980s about how they'd treat an uncomplicated patient in the hospital, and what they'd expect his recovery to look like. This legwork, which began before Doyle came to Milliman, was quite informal. When he showed his prototype guidelines to Milliman, he didn't include substantial documentation, according to court testimony by former Milliman principal David Axene.

As the 1990s progressed, Milliman shifted away from a consensus methodology to one that's billed as evidence-based. The first edition of guidelines to include references was Pediatric HSIM (Health Status Improvement and Management), published in December 1998. Now all guidelines come with references and an annotated bibliography, compiled by a squad of Milliman clinicians, who numbered a dozen as of late 2000.

The lengthy lists of journal articles and textbooks in the guidelines look impressive. All in all, they clearly document a movement from hospital care to care in other settings. But do they justify the specific lengths of stay recommended by Milliman—one day here, three days there?

We asked some of the quoted researchers what they thought.

"These guidelines wouldn't hold up in court"

Milliman's guideline for chronic obstructive pulmonary disease says hospitals and physicians should aim for a two-day stay. The evidence? The guideline cites standards for diagnosing and caring for COPD patients issued in 1995 by the American Thoracic Society. Boston pulmonologist Bartolome Celli, who chaired the committee that prepared the standards, says Milliman shouldn't list the society's work as proof, because it contradicts the guideline.

"We didn't set up a number of days, because we don't know—there's not enough evidence," says Celli, chief of pulmonary medicine at St. Elizabeth's Medical Center in Boston. The published standard made that very point—emphatically: "Insufficient clinical data exist to establish the duration of hospitalization in individual patients to achieve maximum benefit."

For pediatric bacterial meningitis, Milliman lists six articles to justify a goal of three days. The authors of one article, pediatric infectious disease specialist Mobeen Rathore and pediatrician James Waler, both in Jacksonville, FL, say they find this evidence unconvincing. Their own article recommends a six-day stay and discharge only if the patient has been afebrile for at least 24 hours, a criterion missing from the Milliman guideline, but included in the bibliography.

"Bacterial meningitis is a very serious illness, and it triggers a lot of malpractice suits," says Rathore, a professor at the University of Florida Health Science Center Jacksonville. "These guidelines wouldn't hold up in court."

Milliman sets a goal of one hospital day for heart failure and points to guidelines issued in 1995 by the American College of Cardiology and the American Heart Association. Cincinnati cardiologist Laura Wexler, who served on a committee that prepared the ACC/AHA document, wonders why. "We didn't talk about length of stay, so I'm not sure why they quoted us," she says.

Wexler, who teaches at the University of Cincinnati College of Medicine, calls Milliman's recovery path unrealistically short, especially concerning edema. The guideline states that on Day Two—discharge day—the heart patient's pulmonary edema is resolved, and his peripheral edema improved. Wexler argues that it usually takes more than a day to make any real progress with peripheral volume overload, unless you diurese the patient so rapidly that you risk electrolyte imbalances and kidney failure. "These patients might be breathing better after one day," she says, "but if you send them home, they'll be back at the hospital a week later."

As with many other conditions, Milliman also justifies its CHF length of stay by referring to a database of discharge information for HMO patients (non-Medicare and non-Medicaid) maintained by the state of California. Seventeen percent of those with heart failure, says Milliman, were discharged in one day. All deaths and transfers to or from other acute facilities were excluded.

Nashville cardiologist Javed Butler, whose research is quoted in Milliman's heart-failure guideline, finds this database evidence inconclusive because it reveals nothing about the quality of care that anybody received. "What was the readmission rate of those patients discharged after one or two days?" asks Butler, who teaches at the Vanderbilt University School of Medicine. "If most of them came back, we're not achieving much."

Betting on managed care—and getting burned

Guideline experts admit that the amount of published research on hospital length of stay is paltry. But that doesn't mean guidelines still can't offer sound advice, say Milliman sympathizers.

"Eighty percent of what we do in medicine lacks scientific backup," says oncologist Lee Newcomer, a former vice president of health policy for Minneapolis-based UnitedHealth Group. Now an executive for a health care dot-com called Vivius, Newcomer recalls how a few of his mastectomy patients in the 1980s didn't want to spend a night in the hospital. "Our surgeons were able to do that for them," says Newcomer. "We didn't have time to write it up in a journal, though."

That kind of anecdotal evidence gave rise to Milliman's notorious same-day mastectomy guideline, which Newcomer views as unfairly impugned. "A same-day mastectomy is a doable procedure of choice for some women," says Newcomer, "but you wouldn't want to demand it for everyone."

Says internist David Nash, director of the Office of Health Policy and Clinical Outcomes at Thomas Jefferson University Hospital in Philadelphia, "Milliman's negative reputation is undeserved. The criticism is just a smoke screen for a bigger problem, and that's the inefficiency of medical care. Some people do stay in the hospital longer than they need to."

No matter what reputation it deserves, Milliman's guideline business has struggled lately, racking up an $8 million deficit as of March, according to a deposition by a former executive. As managed care has retreated in the face of withering condemnation, so have efforts to institute efficiency through guidelines and the like. That's one reason why the Milliman guidelines have been on the auction block, says Minneapolis internist and medical-software entrepreneur Solomon Zak. (A spokesman for Milliman denies that's so.)

"Their market is going away from them," says Zak, who says Milliman offered to sell him its guideline business. (He declined.) "Our society doesn't want to be held accountable for the health care it consumes."

Consultant Robert Bruckman agrees. "All guideline writers are losing business, because in the backlash against managed care, insurers aren't as strict as they used to be. They won't tighten up again until costs get out of hand."

The failure of capitation to become the standard source of payment for health care also bodes ill for Milliman, says cardiologist Richard Vernick, a consultant with The Hunter Group in St. Petersburg, FL. The theory behind capitation is that if you pay an integrated health system up front to care for patients, it'll invest some of that money in preventive care, home health services, and subacute facilities, all of which should reduce hospitalization. And Milliman guidelines assume that health systems offer such extensive services.

"But with the pendulum swinging back to fee-for-service, there's less of a financial incentive to develop this kind of network," says Vernick.

Of course, back in the mid-1990s, David Axene predicted a future dominated by capitation and HMOs. That's why hospitals would become ghost towns.

It wouldn't be the first time that someone bet on managed care and lost.

Milliman's Texas lawsuit: Did a "bad marriage" produce bad guidelines?

The story of Milliman USA's partnership with the University of Texas-Houston Medical School to develop pediatric inpatient guidelines sounds like the echoes of a bad marriage, in which phrases such as "I assumed" and "I thought we agreed" replace clear communication.

Now everything is being slowly and painfully spelled out in court. Four UT pediatricians have sued Milliman and a fellow faculty member who headed the project for listing them as guideline contributors without permission. They claim that their association with the 1998 edition of the guidelines—which they call "a disreputable commercial product"—has hurt their reputations. No matter how the suit turns out, Milliman's reputation has arguably suffered, too.

In the beginning, it seemed like a good match. Milliman thought that the UT pediatric department would enhance the guidelines' credibility, according to James Turner, former executive vice president of the guideline division. In return, UT would receive $100,000.

The project's point person was UT pediatrician Robert Yetman, an advocate of efficient care and a spirited visionary. "He's a 'wouldn't it be nice?' kind of guy," says one colleague. Yetman declined to be interviewed.

Up to this time, Milliman's pediatric guidelines were scattered throughout several miscellaneous volumes. Milliman recruited Yetman in the mid-1990s to assemble these guidelines in a single volume, update them, and develop new ones. Yetman, paid separately by Milliman as an independent contractor, brought aboard UT colleagues to fine-tune the material, especially in fields outside his expertise. One colleague was pediatric infectious disease specialist Thomas Cleary, as cautious as Yetman was bold.

When Cleary reviewed Yetman's drafts, he didn't like what he read. A goal of three days in the hospital for endocarditis? "There may be rare patients ready for discharge at seven days—three would be inappropriate," Cleary wrote on the draft. Three days for bacterial meningitis? "Rarely appropriate." To him, these illnesses were too life-threatening to trifle with in the name of shorter hospital stays.

Cleary's concerns weren't lost on Yetman. When The Wall Street Journal interviewed Yetman in April 1998 for an article on Milliman, Yetman suggested that the reporter also talk to Cleary "for opposing views."

Cleary says he discussed his misgivings with Yetman in both the spring and early summer of 1998. "I pointed out that . . . children were going to die because of what he had written, and he needed to change it," Cleary said in deposition testimony. One sticking point was the Milliman "day" in the hospital. Milliman defines it as a phase—it may be less than 24 hours, or more than that, depending on the patient's progress—but Cleary warned Yetman that insurers would stick to the normal definition of day to deny needed care.

Yetman saw things differently. He says Cleary had expressed fears not about the guidelines themselves, but how insurers would abuse them. But then, Cleary was confused about how they worked. Yetman explained that the guidelines contained safeguards—that you couldn't discharge someone to home health care if it wasn't available, for example—and reassured Cleary that he'd remain in clinical control. Yetman thought he had made a believer out of Cleary.

"The first time . . . I actually heard from his mouth that he thought the guidelines were inappropriate was in . . . October [1998]," Yetman testified.

Meanwhile, Milliman was receiving less than glowing reviews from the American Academy of Pediatrics, which also scrutinized a draft of the guidelines. "A clear consensus from our reviewers is that many of these pediatric inpatient guidelines are consistent with their clinical experience, but that many others are not," an academy letter stated. "The guidelines too often assumed rapidly reversible conditions."

Milliman finally published the guidelines in December 1998, but Cleary didn't see the book until September 1999. It listed him and 16 other UT faculty members as contributors. The guidelines he considered dangerous were still there.

Cleary was livid. He says that he hadn't expected to see his name in the book, nor had he authorized anyone to use his name. Based on everything he had heard, it was Yetman's book.

Yetman thought he had permission to list the names of contributors because they had participated in the project, and enthusiastically so. "If they work on it, you would obviously put their names on it," he testified. "It is something that you would expect to do in an academic environment of friends. There was no written consent."

Cleary demanded that Milliman remove his name. He quotes Yetman as replying that Milliman would do so in the 2000 edition. Incensed by the lack of immediate action, Cleary sued both Milliman and Yetman. "The only way I can repair my damaged reputation is . . . to scream . . . from the hilltops, if need be," Cleary testified.

His fellow plaintiff was pediatric endocrinologist William Riley, another listed contributor, who left the UT faculty in 1997. Riley objected to a Milliman guideline for diabetic ketoacidosis that aimed for a one-day hospital stay. Earlier this year, two more listed contributors, UT professor Steven Culbert and Steven Wolfe, now practicing in Arizona, joined the suit.

So what are the plaintiffs hoping for? At one point in settlement talks, they wanted to stop Milliman from publishing pediatric length-of-stay goals. A recent court filing by the plaintiffs, though, merely demands that Milliman recall all volumes of the 1998 edition, as well as pay unspecified damages. Milliman's Web site says guideline purchasers will receive a revised list of contributing authors.

William Riley says he's not an enemy of practice guidelines per se. "But Milliman's are designed to get people out of the hospital quickly to save money for insurance companies," says Riley, now vice president of medical education for Driscoll Children's Hospital in Corpus Christi, TX. He's afraid that the diabetic ketoacidosis guideline, if enforced for all patients, could "kill people."

Cleary harbors the same fears. "Milliman's guidelines reflect little understanding of medicine," he says. "In some cases, they'd force doctors to commit malpractice."

Robert Yetman states his opposing opinion just as strongly. He thinks the UT pediatric department should be proud of the guidelines. The controversy, though, has taught him one lesson: The next time he collaborates with other doctors on a writing project, he notes, he's probably going to get written permission to list their names as contributors.

While legal squabbling fills the air, another UT pediatrician who declined to be identified says the guideline saga doesn't have any villains, only casualties.

"They're all doctors," says the colleague, "with a great desire to give good care to kids."


*To be covered in an upcoming issue of Medical Economics.

Robert Lowes. Hospital-stay guidelines: Just plain weird. Medical Economics 2001;15:86.