Feature Story

Turning state’s evidence

Diabetes study protocol
Registry reports

July 2001
Mark Uehling

Vermont labs are funneling critical diabetes data directly to clinicians, thanks to a far-reaching program.

NASA has warehouses full of computer tapes. These unmined, unexamined Himalayas of data are collected from decades of space missions. Is the typical HIS or LIS all that different?

Of course not. Like the NASA warehouses, laboratory and hospital information systems contain idle information—hints of overlooked ailments, test values that a physician probably glanced at but never acted on.

That may be changing in Vermont, where computers have been programmed to give primary care physicians a one-page summary of diabetes-related laboratory data. The sheet includes the patient’s most recent HbA1c, microalbumin, and lipid panel, and it summarizes the last four results of each of these tests.

"All the diabetes information is right there in one place," says pleased Vermont internist Charles MacLean, MD, of Fletcher Allen Health Care, the largest provider in the state. "It saves me time and hassle in getting what
I want."

Dr. MacLean, who helped launch the program, says the same diabetes information is also periodically commingled from all his patients to allow him to see how they have been faring under his care. "We can crank out these population-based reports to our docs and say here are your numbers and here are the numbers for your practice. Here are the numbers for the whole organization, and here are some national benchmarks. You can really say, ’Hmm, those guys are doing better than me; what could I be
doing differently?’"

And the computer does not just tally the patients who kept their physician appointments. It also tracks diabetics who did not show up for testing. "You’re really taking care of an entire population of people, not just the ones who happen to come in," says Dr. MacLean.

The changes in Vermont would not have been possible without the computer, but the human element was critical. As Dr. MacLean notes, winning over the staff was a challenge. "This didn’t take a whole lot of programming skills. The biggest part of it is the process, the cultural change. How do you identify the people who are going to print it out in the office? How are you going to service the docs? That process takes a lot of time and discussion—and people championing it."

From the beginning, the idea in Vermont was to weave together the people and the computers. In the context of diabetes, that meant helping physicians look at patients not only individually, but as a population with a shared disease. "It’s not often that we take a step back and say, ’Okay, I’m taking care of thousands of people with hypertension—how am I doing?’" Dr. MacLean says. "’Am I getting most of them under control?’"

The Vermonters, to be clear, did not just build a diabetes registry. Commercial software can handle that. Rather, the team in Burlington designed a diabetes registry that would gather HbA1c results and other data over months and years and funnel it to primary care physicians.

Diabetes registries that gather and disseminate results are common. Diabetes registries that win support are rare. The challenge is less technical than it is institutional, political, tactical. Dr. MacLean reports he and his colleagues like the system and have no complaints about it.

"If the doctors are happy, that’s wonderful. That tells you it’s working," says endocrinologist David Goldstein, MD, professor of pathology at the University of Missouri. Dr. Goldstein is also an investigator with the renowned Diabetes Control and Complications Trial and chair of the steering committee of the National Glycohemoglobin Standardization Program.

"Programs like Vermont’s will help educate physicians and patients about achieving optimal diabetes care," Dr. Goldstein says. "Vermont is leading the way in what has been a national movement to use HbA1c testing in patients with diabetes to assess the quality of care."

As Dr. Goldstein points out, his own institution has had a diabetes registry for a decade, and a database in Washington’s Puget Sound region contains records on 15,000 diabetics. But the greater reach of the Fletcher Allen registry—which includes anyone with an ICD9 code for diabetes—appears to be novel. "We don’t have a system that encompasses all people at the system who are not registered at the diabetes center. If all such patients are registered into the Vermont system, that’s good," Dr. Goldstein says.

Dr. Goldstein also mentions another wrinkle: quality control. Most hospitals are obligated to report some HbA1c statistics to the federal government. Under the Health Employers Data Information Set, or HEDIS, hospitals must track the prevalence of HbA1c and the percentage of patients with HbA1c test results greater than 9.5 percent in measuring the overall quality of their care. So there’s a payoff to registries like Vermont’s beyond the benefit to patients.

Even so, Dr. Goldstein says, "The Vermont people should be congratulated on working hard at educating people about diabetes care. Lab people are getting pretty savvy. Clinicians have a ways to go. The lab people need to work closely with clinical people, teaching them about this test and interpreting what the test results mean."

Interpreting the tests is easier if the laboratory in question is participating in the CAP Survey for HbA1c, which Dr. Goldstein praises lavishly. Also important: using a method certified by the National Glycohemoglobin Standardization Program.

The Vermont program met both of these criteria, attests pathologist Mary Tang, MD. She consulted briefly with the information systems staff about instrument selection "to make sure the hospitals in the state are using DCCT-approved analyzers, so that we would have a standardized reference range, and so that we would conform with the American Diabetes Association [ADA] guidelines," Dr. Tang says.

HbA1c testing volume has risen sharply in the Fletcher Allen network. Dr. Tang suspects some of the increase in volume could be linked to equipment-related difficulties at other laboratories in the state, which then turned to Fletcher Allen’s laboratory in Burlington. But Dr. Tang also credits the ADA with raising awareness of HbA1c testing, chiefly by promoting its own guidelines. "We are getting more people tested, one way or another," Dr. Tang says.

You can say that again. One of Dr. Tang’s colleagues, Michael L. Gagnon, director of eHealth and pathology informatics in information services, and formerly director of laboratory business development, at Fletcher Allen, notes that from March 1997 to March 2001, HbA1c testing increased 54 percent to 28,080 tests annually. As a rule of thumb, Gagnon says, the population of diagnosed diabetics rises by a single percentage point a year. "The remainder we attribute to increases in testing on known diabetics and an increase in the number of previously undiagnosed cases," he adds.

Some of the best indications of improvement in the patients in the registry are under wraps for now, as the team in Vermont submits the work for publication in a peer-reviewed journal. But in a pilot study of the registry, 66 percent of Fletcher Allen’s diabetics had at least one HbA1c in a year, and that percentage later rose to 81 percent.

Since then, more of the state’s diabetic patients have had ophthalmology visits every year, and more have had at least one microalbumin run annually. Patients in one Fletcher Allen pilot study were more likely to have an HbA1c result of less than eight percent than those not enrolled. The average HbA1c value has gone down. And all of the results are statistically significant. That’s admirable progress for tweaking a few databases and ensuring that the right piece of lab-based paper meets the primary care physician’s gaze at the right moment.

So how did the Vermont team make it happen? Parts of the story can be replicated elsewhere, and others are unique to Vermont. As Gagnon jokingly told an audience at the 2001 Executive War College, sponsored by The Dark Report and held in May, consumers and businesses in the Green Mountain State prefer homegrown products like Ben & Jerry’s ice cream.

The strong local bias is relevant to health care because the desire to avert a hypothetical invasion of non-Vermont businesses, whether dairies or laboratories, is widespread in Vermont and need not be articulated. It’s part of the Zeitgeist.

This is why Fletcher Allen has such amicable relations with 14 laboratories around the state in the North East Community Laboratory Alliance. The NECLA consortium was able to study all of the HbA1c findings in the state. Just doing the study seemed to increase the use of HbA1c testing, and that atmosphere of trust went a long way. "It’s one thing to collect the data from one hospital, but it’s another to collect it from multiple hospitals," Gagnon says.

The same collegial atmosphere applied to building the diabetes registry. It was not simply a matter of one lab’s data being pumped into a physician’s office. It involved figuring out how to guarantee that a busy physician would see that information once it was in the system. It involved talking to renal specialists and vascular surgeons about their diabetes patients. And as with change in any industry, people resisted at first.

In Vermont, Gagnon says, part of the way to quell the doubts was with reports that internists and family practitioners had the option of using. Reaching out in a low-key, noncoercive way mattered.

Gagnon explains: "The physicians get their rosters, and it warns them when patients have not been in. It says, ’There are no results on this patient in the last nine months or a year.’" At that point, the Fletcher Allen system swings into action. It may encourage a doctor or practice not to let a particular patient slip through the crack: "We’re on the phone. Fletcher Allen physicians are called. We’re saying, in a friendly way, ’Do you know that [Jane Doe has not had an HbA1c in 18 months]? And here’s the data.’ It’s a carrot rather than a stick approach," says Gagnon.

That the project would succeed was clear almost immediately. The project quickly expanded from a pilot stage (with two physician groups) to where the chief medical officer wanted the diabetes summary sheets applied to all of the medical center’s 6,500 diabetics. As Gagnon reports: "After the first pass of data collection the first year, suddenly everyone wanted something quite different. They recognized they could get a lot more out of this than what we had anticipated." Resistance dissipated; momentum grew.

Now, Gagnon says, a variety of nonlab, nonpathology types at Fletcher Allen believe there is potential in expanding the role of the laboratory. "We talk about it a lot at these national meetings, saying, ’Well, the lab holds all this data, and if anyone paid attention to us, they’d start to realize what we can contribute.’ People are realizing what we can contribute. In this case at least, we’ve jumped over that barrier. People are starting to recognize the value of this."

Putting the theoretical value of HbA1c into ordinary clinical practice is something else that Vermont may be pioneering. In April, the American Medical Association, Joint Commission on Accreditation of Healthcare Organizations, and National Committee for Quality Assurance announced they will encourage more regular HbA1c testing.

HbA1c is not without controversy, especially as a tool to identify undiagnosed cases of diabetes. But there is little argument that as a cumulative indicator, the test can help a primary care physician determine how aggressively to intervene. That’s because HbA1c values are retrospective, long-term numbers; they are a numerical gauge of a patient’s glycemia over the 60 to 90 days before the test.

At a cost of $100 billion, with 8,000 new cases of blindness, 120,000 nontraumatic amputations, and 160,000 deaths annually, diabetes mellitus is a monumental disease-management challenge. And it’s growing. The incidence jumped 33 percent during the 1990s, according to the Centers for Disease Control and Prevention.

The problem is that laboratory information to stop the disease from progressing doesn’t reach the physician at the right moment. "Providers want to provide the best care, but often they don’t know—particularly in urgent visits—that they’re seeing a diabetic," says Susan K. Levinsky, RN, MPH, clinical improvement coordinator at Fletcher Allen. "You just see the patient in front of you."

Levinsky acknowledges that Fletcher Allen’s Gagnon saw things differently from the start. "If it weren’t for the involvement of Mike Gagnon and the lab, this wouldn’t have gotten off the ground," she says. "Having the lab person understand that there’s a bigger picture, that population information is clinically useful, as opposed to just one lab value at a time, was key."

As Levinsky makes clear, the laboratory and the way it handled information cast a spotlight on a much lamented fact of diabetes: Patients don’t always look after themselves. Once diabetics are identified at Fletcher Allen, a variety of clinical appointments are scheduled automatically. "But if they didn’t come, they didn’t come. We had no systematic way to make sure this was happening," says Levinsky.

As always, some patients vanished, only to require amputations, dialysis, and more extensive care at a later date. Which was not what HMOs or the quality improvement experts in Vermont’s Program for Quality in Health Care would prefer. Among their goals and recommendations was to have every diabetic get an HbA1c at least two and as many as four times per year. That was not consistently the case for all diabetics in Vermont or, for that matter, anywhere in the nation.

Inspired by guidance from the state of Vermont, Levinsky, Gagnon, Dr. MacLean, and a team from Fletcher Allen’s information services department devised a way to automatically generate cover sheets for diabetic patients in two large primary care practices affiliated with Fletcher Allen. Says Levinsky: "The average chart is two inches thick. This is a quick and easy way for physicians to be more efficient with the data that is already available to them. It’s not really new information; it’s just presented in a new format."

Levinsky would be the first to concede that what she and the Fletcher Allen team have accomplished is no more than a basic piece of paper containing demographic and diabetes-specific data. "It has the date and level of the most recent HbA1c," she says. "It’s a reminder clueing in the physicians and providers that you have a diabetic standing in front of you—you should address the diabetes care."

If the simplicity of the sheet was an asset, so was the freedom that accompanied it. In the past, HMOs have attempted to strong-arm primary care physicians into shepherding diabetic patients into the office. That typically has failed. "We just gave them the information," says Levinsky. "We hoped that when they saw the patients, they would address it. Some of our physicians called up our patients and said, ’Gee, Mr. Smith, you haven’t been seen in a while and how’s it going? Do you think you might want to come in?’ That went over well."

So did amassing the laboratory’s data for each physician. Every Fletcher Allen internist or family practice physician got a roster of his or her 40 to 50 diabetic patients. Says Levinsky: "Most physicians didn’t have any idea of the number of their diabetic patients. Most, at first blush, thought they had a lot more than that. When they realized it was only 40 to 50 patients, they felt like it was a manageable number. With that list in front of them, they could do a chart review, or they could just look at appointments and see if patients had been in."

At that point, many physicians got involved in winnowing out patients who had died or moved. That left a shorter list of patients who needed their attention.

Rethinking how to present the laboratory’s existing data has made a difference for Burlington’s diabetics. "We certainly identified patients who were diabetics and hadn’t been to the doctors in years," Levinsky says. "We were able to address the diabetics who were diagnosed but never came back for followup care."

The only bump in the road, she admits, has been some patients’ reaction to being in a database. "We’ve had patients say, ’What do you mean, I’m on a registry?’" Levinsky says. The privacy issue is not trivial. Social stigma and workplace discrimination can follow any diagnosis of diabetes. "They don’t like the thought that someone other than their primary care physician might have access to their clinical care information," she says. "But we’ve been very careful about any time we share statistics [for research], there is never patient identification information."

In the information systems department, meanwhile, Gagnon was promoted from the laboratory to overseeing more global projects and budgets. But he has not forgotten the diabetes project, and he speaks of its next growth phase with the same fervor as a dot.com entrepreneur half his age. Managing other diseases using the same techniques could be one natural extension of Fletcher Allen’s efforts and draw just as heavily on the laboratory’s untapped knowledge.

Another possibility, Gagnon says, would be to build a diabetes registry for Vermont, not just Fletcher Allen’s home turf. That would seem eminently possible because of the North East Community Laboratory Alliance. Says Gagnon: "The situation here in Vermont is quite unique. Because virtually every hospital does the vast majority of their testing for their physicians, we collect virtually every lab result in the state on everybody. That is a really powerful concept. We’re going to try to be the model for this."

Then Gagnon launches into the details, mentioning, among other topics, the Logical Observation Identifier Names and Codes, or LOINC, cross-reference codes; master patient, provider, and procedure indices; and how to find hundreds of thousands of dollars in funding. It becomes clear that a few hurdles remain. But that’s fine. Gagnon’s overcome them before.

Mark Uehling is a freelance writer in Chicago.