How can a single corporate employer afford to invest so much effort in improving quality in managed care, knowing that its pioneering work may simply be copied?
As manager of international compensation and benefits for Xerox Corp., Helen Darling directs the purchasing of health benefits for more than 144,000 employees, dependents and retirees, and oversees contracts with more than 200 HMOs. She also has overall management responsibility for implementing Xerox's health care strategy — called "HealthLink" — and for managing the company's fee-for-service insurance.
But Darling's influence on the evolving shape of managed care goes beyond the halls of Xerox. She co-chairs the National Committee on Quality Assurance's Committee on Performance Measurement, which oversees the evolution of HEDIS, the Health Plan Employer Data and Information Set. NCQA and HEDIS are setting the standard in managed care accreditation. In contracting with HMOs, a growing number of national employers require NCQA accreditation, or real progress toward it.
Darling is a past president of the board of the employer-created Managed Health Care Association, and a member of the National Advisory Council of the Agency for Health Care Policy and Research (AHCPR).
Before joining Xerox in 1992, she was a principal in the health care consulting practice of William M. Mercer Inc., specializing in cost containment, plan design, Medicare, retiree medical benefits and national and state health policy. Darling has been a health researcher at the Institute of Medicine of the National Academy of Sciences, the U.S. Department of Health and Human Services and Rhode Island's Health Services Research Department. In the 1980s she was senior health policy and health benefits adviser to U.S. Sen. David Durenberger, the ranking Republican on the health subcommittee of the Senate Finance Committee.
Recently, Darling spoke with Managed Care's senior contributing editor, Patrick Mullen.
MANAGED CARE: What are some of the experiences you've had with the health care system that inform how you do your job and shape the expectations you bring to your work?
HELEN DARLING: Probably the most significant impact on my thinking has been that I have had experience and some training as what is termed a health services researcher, doing the kind of work that today is embodied in the Agency for Health Care Policy and Research. As a consequence, I carry around in my head, with varying degrees of specificity, beliefs about the system that have been informed by health services research.
MC: What would be an example of that?
DARLING: A really good example would be that health services research has documented the very strong positive relationship between volume and quality, whether it was in surgery or various treatments. For example if you are going to have open heart surgery you want to find the hospital and surgeons that have done the most open heart surgery. If you're having a high-risk pregnancy, you want to get to the hospitals that specialize in high-risk babies as soon as possible — even if you're about to lose the baby. There's good research showing that if you can get people into the hospital or in a facility near the hospital and give them bed rest and keep the baby in utero as long as possible, there will be a significantly better outcome. So, there is just example after example.
MC: Why did you come to work at Xerox?
DARLING: The company's approach to health care was philosophically consistent with what encourages people to select organized delivery systems. These systems are organized around a whole bunch of things: quality health care, performance measurement, member satisfaction. It makes a lot of sense to me and it's the sort of thing I enjoy working on. The other piece is that I had certainly long felt that we ought to say to people, "Here's the fixed allowance that we will pay for health benefits, whether in the Medicare program or the Medicaid program or at a large employer or for state or federal employees. If you want to do anything differently, like stay in a fee-for-service plan, then you pay the difference. Fortunately, Xerox was able and willing to give a benefit allowance that actually pays more than 100 percent of the cost of an efficient HMO, with a very comprehensive benefit package.
MC: One of the things that has marked Xerox's approach is the confidence you've placed in the National Committee for Quality Assurance, NCQA. I understand that the company pays a $120-per-year incentive to employees who join NCQA-accredited plans, and you freeze enrollment in plans that are not moving toward accreditation. What drove those decisions?
DARLING: A couple of things. First, we were absolutely committed to quality as a company across the board. The governing principles of our purchasing include what we call leadership through quality. It's how we buy everything and it's how we do everything. The message we want to send is that we believe plans that successfully achieve external accreditation bring a special value that we want to reward. We also recognize that it costs them money to become accredited. We want to let them know that we care so much about meeting those standards that we will make it a little bit easier on them by giving them a financial advantage. Those plans are going to get more of our people because of that financial advantage, and that's exactly what we want to achieve.
MC: How would you respond to the folks who say that NCQA measures some of the wrong things or that it's moving too slowly toward measuring outcomes?
DARLING: Well, I have to say that I'm certainly not unbiased, but I have kind of a rage reaction. First of all, if anybody walking the Earth looks at HEDIS 3.0 and thinks they can do a better job of finding viable measures that are meaningful, relevant, scientifically valid and reliable, we would welcome their recommendation. Anybody who criticized them has never been able to come forward with better substitutes. The best brains in the country developed those measures. We have Rand Corp., Harvard, AHCPR, the Centers for Disease Control and Prevention. We have a brain trust working on performance measurement for health plans. There isn't anything better than that, so that's partly why I have a kind of a rage reaction.
MC: The Joint Commission on Accreditation of Healthcare Organizations has been working to re-establish itself in managed care accreditation as an alternative to NCQA. Does it make sense to have more than one accrediting body?
DARLING: Certainly there's plenty of work for everybody to do. What's important is that we have, at a minimum, very high standards, and that all health plans and everybody who takes care of people meets those high standards. We have chosen to support NCQA for a long time. It was an easy choice at the time, because they were the only ones doing it. The other thing is that we would distinguish between real managed care and organized delivery systems. This isn't a criticism of the Joint Commission; I'm a long-standing colleague and friend of Dennis O'Leary, and I have a great deal of respect for him and his team. But long ago, when I was involved as a health services researcher, I found that hospitals and doctors on average, including the people who accredited and inspected and analyzed those institutions, tended to either be anti-managed care or simply not informed about it. So if the Joint Commission becomes much more heavily involved in real managed care accreditation — and Dennis has heard me say this — I still think there will be plenty of room for a second accrediting body. I doubt there's room for six or seven, especially when we're trying to standardize. One of the biggest concerns is that if we move away from real standardization we've got serious problems, because we've got enough of a struggle to get everybody agreeing on and meeting the same standards. If we change them or water them down, we're going to make it very hard for both purchasers and consumers to make decisions.
MC: That's a concern with HEDIS that I've heard from some people on the health plan side, who say that as soon as they get used to a set of measures, it changes.
DARLING: That's not true. First of all, HEDIS 3.0 was the first generation of real measurements that had the level of detailed specificity and scientific validity that is acceptable. I'm one of the main authors of HEDIS 2.0 and 2.5, and we did the best we could at that time under incredibly tough circumstances, but HEDIS 3.0 was in a whole different league. But we didn't throw out all of HEDIS 2.0 and 2.5. What we said is, we're going to evolve this system. We don't want people setting up systems that are frozen in time because we want to get better every year. I would guess that if you looked at HEDIS 2.0 and 2.5, which really should be looked at together, and compared them to HEDIS 3.0, the required reporting set contains more than 80 percent of the same measures. We want continuous improvement and the number of new or significantly changed measures will be quite small. By the way, we're no longer going to use version numbers like 3.0; instead, we'll name each version for the year it's issued, like HEDIS '98. We want to send the signal that what we're making are small, incremental changes.
MC: Will 1998 will be the next stage?
DARLING: What we want to do is continuously refine and evolve the performance measurement system, including taking measures out. We want to reserve the right to change enough so that we're really detecting what's going on. I know that HEDIS critics worry that the plans will just pay attention to the things that are measured and somehow ignore the things that aren't. Well that's just too cynical, even for me.
MC: It presupposes that it would be possible in practice. How could you get a whole organization to meet standards on that selective a basis?
DARLING: Life is just too complicated for that. You're absolutely right. But also what we're trying to do is say we want to measure the quality of the performance and we would like to be able to rotate measures, to put some back on and take some off without creating chaos for the plans. We know that the HEDIS measures give us just the tip of the iceberg. We like to think that it's an important tip, and that these measures in fact statistically lend themselves to plan comparisons. That's different from quality improvement within a plan.
MC: How can HEDIS be improved?
DARLING: The problem is not that there's stuff out there that we're not doing. Improvement is needed in two broad areas of problems in health plan performance measurement. The first is that there is in many instances a highly tenuous link between any medical intervention and any health outcome. The second is all the statistical issues. Even if there is a link between interventions and outcomes, it may take twenty to thirty years and an awful lot of cases before anything can be measured validly. For example, it takes not only five years' worth of survival to get five-year survival rates, if that's what you want. You also have to have hundred of thousands of members to have enough cases. Thank God, even though something like breast cancer seems common, the number of breast cancer cases is actually statistically quite small in a given health plan or any given group of people. Say you're trying to claim that Plan A is better than Plan B at breast cancer care and you want to ignore things like mammograms and biopsies and just talk about survival rates. It would take ten or fifteen years to reliably and validly determine whether medical interventions made a difference. So what we're doing is what's possible statistically and scientifically. What we are not doing is being irresponsible in what we're putting forward. And some of the people who criticize it, frankly, are either being disingenuous or don't know what they're talking about.
MC: Things would seem to be complicated by the fact that people don't stay in the same health plan fifteen or twenty years — or even five or six.
DARLING: Actually, there's a lot less moving than people think. It's still a problem, but the day will come when we can actually link data bases among plans. On the other hand, I also know that there's a lot of interest in stopping access to data. Some of the well-meaning patient protection and confidentiality rules that have been proposed would, in fact, kill research. Now, I understand why some people are worried, and I certainly think that patients as individuals need to be protected. But we can protect people so well that research in this country comes to a grinding halt.
MC: Solving the statistical problems would seem to be a matter of refining techniques over time.
DARLING: Well, yes and no. The sample size problem is insoluble without making all plans about the size of Kaiser in Northern California. Recently, I was the purchaser representative in a discussion among a group of smart people who are working to refine these measures. They made an interesting point about the sample size problem that I think is exciting. Let's say the average person is less concerned about how a plan handles breast cancer or prostate cancer, but wants to know whether they and their family will be taken care of in the right way when they're sick. Even though I know my genetic code pretty well, in the sense that I know what my relatives died of, I can't tell you whether I'm going to have heart disease, cancer, diabetes or whatever in my next 30—40 years. So what I really want to know is whether my health plan is going to handle me perfectly, whatever might go wrong with me. Now if that's true, then one of the ways we might solve the problem of sample size is by putting together what we would call a family of diseases. Let's say we come up with five to 10 measures having to do with heart disease. We would look at every one of those, and we'd have a smaller sample size for each one, but we would lump them together. Then we could say, based on the aggregate data, there is evidence that this plan is better at heart disease care than another plan. Or all cancers could be combined, for example. I was told by the people sitting in the room, who know about these things far more than I, that this actually could work. So that's the kind of methodological refinement that we're all busy working on right now.
MC: So the challenge to some degree is political, getting people to share data?
DARLING: No, I'm still talking about within the same plan. Let's say we're talking about M.D. Health Plan in the Washington, D.C., area. If I ask them about breast cancer, they're only going to have maybe five cases in a whole year. But if I ask about seven different kinds of cancer, of which they might have 35 cases, I'm going to get more information. I'll get a score out of that telling me if the plan is proficient with all of those or only some of them.
MC: A number of legislative bills are floating around that would require plans to present information to members in so-called "plain English." For a few years Xerox has distributed plan summaries in a common format. What lessons can be drawn from your experience about the difficulties of taking relatively complex information and making it understandable?
DARLING: I think people understand measures of plan performance if you just tell them what they mean and why they matter. An example is physician turnover rates. There's something possibly wrong at a plan that has high physician turnover. Either it's not doing a good job in recruiting and retaining physicians, or there may be some management turmoil, or it's not treating its physicians very well. All of those are bad news. You can go through each measure, readmission to hospitals after depression or a hospital visit after thirty days. That suggests that they're not doing as good a job taking care of people with mental health problems as they should be. We just put it in really plain English.
MC: And yet is it correct that you're going to discontinue some of the summary information?
DARLING: Yes, but just because of money. It's expensive to print it. Sometimes we can't afford things we know are worthwhile.
MC: That raises the issue of the money Xerox has been spending measuring health plans. Arguably your shareholders might wonder why they're paying for all this work to make HMOs better. Is it work that's going to end up to the benefit of other employers who are not making this investment? Is that a sustainable kind of commitment?
DARLING: You always have to evaluate the trade-offs and the costs, and we've actually done a pretty good job. I probably get asked, and others around the company get asked, maybe once a day to get involved in a group at the National Institutes of Health or Harvard or someplace, because people are going to do wonderful things and they've absolutely got to have a purchaser there. I alone turn down one a day, five days a week, fifty weeks a year, because we have to stay focused on a few things that we think have big payoffs.
MC: Just from reading the list of groups in which you're involved, you say yes often enough. You personally probably are spread fairly thin in terms of the commitments you make.
DARLING: First of all, we are a large very public corporation with a long-standing commitment to a civilized and good society, so there are a lot of things we do that might not be obvious to everyone as being in our narrow corporate self-interest, like helping with campaigns for diabetes, or being involved in "welfare-to-work" programs. We say if the society that we live in isn't good, then what kind of world do we have and what sort of business and what sort of people are we? A subset of that is if we've got contracts with more than two hundred HMOs across the country, that's about 40 percent of all the HMOs. So if we work hard within NCQA and with the Committee on Performance Measurement, and if I serve on the board of the Medical Outcomes Trust or something like that, and all of us working together advance the practice of good health care in the United States, then our people definitely benefit from that. And we're not the only ones doing that. There are other companies such as GTE and Digital that work alongside us. Together we move the system along, and together we benefit.
MC: One hot topic these days is the question of whether having health care as a for-profit industry in some quarters has gone too far, and whether care has sometimes suffered in the interests of profits. For example, there's a lot of controversy — and some litigation — about the for-profit Columbia/HCA hospital chain. What's your thought on that? Can health care work for the benefit of patients as a for-profit business?
DARLING: Whether something is for-profit per se may or may not be important. It varies enormously with the plan. Xerox is a for-profit company, but we never lose sight of the need for quality. We wouldn't exist as a company if we let quality slip, because we've got lots of competitors. It's also true that it doesn't help at all that some of the for-profits have sometimes behaved in ways that could be best characterized as greedy, by saying things like, "Oh, we can't afford to pay for one more day of mental health treatment," and then turning around and crowing about their millions of dollars in profits. There's a level of taste and judgment and a sense of what is a reasonable rate of return that are very important. What I don't like is to see them all lumped together, because there are good for-profits, and while I haven't seen many bad nonprofits, there sure as hell are some really inefficient ones. By the way, I'm in a nonprofit HMO — Kaiser Permanente.
MC: And how has your experience been?
DARLING: I love it, but I've always loved it. I'm speaking only as an individual right now. It pleases me that doctors make all the decisions in my health plan, that physicians are the dominant force, that physicians determine what they do and don't do. If they have to figure out what to do next, they're not doing it worrying about five million institutional shareholders. They don't have to return rapidly growing earnings. The trouble is really not the for-profits, it's that they are tied to the volatile stock market and the pressure to achieve short-term quarterly earnings. I think it's worrisome that people have to sit around and worry about whether they're going to generate enough earnings that quarter to please a group of stockholders who are looking for fifteen- to twenty-percent return on equity.
MC: Kaiser and similar plans are going through a transition as the model they represent continues to face competition from a whole slew of different kinds of managed care. Do you think it's a viable model in the long run? To what degree do they have to change?
DARLING: Well, I think they need changing a lot. I happen to like the model. It's my favorite, but I'm a realist. I regret that other people in America don't love it as much as I do.
MC: There was once a consensus that eventually we'd all be in group- and staff-model HMOs. That clearly didn't happen. Why not?
DARLING: There are a couple reasons for that, which are a little bit unrelated to the reason people thought that's where we'd be. One is the sheer cost of that model, the huge capital investment needed to make that happen. And that's unsustainable over the long haul. The second thing is that we were basing those ideas on the fact that these group- and staff-model HMOs operated in certain comparatively narrow markets, by U.S. standards, where people were very comfortable with that type of health plan. An example is Group Health Cooperative of Puget Sound. You have a different culture, and people just behave that way out there, whereas in other markets they do not. I do think Kaiser already has changed a lot. Interestingly, two of our most successful markets, where Kaiser is doing very, very well with Xerox people, are the Washington, D.C., area and Atlanta. I was really surprised by that. I never would have thought those would have worked in those markets. Yet they're booming in Atlanta and outside Washington.
MC: Let me ask you one last question. With the failure of the Clinton Plan of 1993, it seems that at least for the time being, we've decided as a society that we're going to keep the system we've got, which is largely employer-sponsored health care plus Medicare and Medicaid, and see what we can do to improve that, rather than go to another kind of system. Does that make sense? If you were designing a health care system, would that be how you'd go about it?
DARLING: I don't think it's worth speculating about. It will never happen in our lifetimes. I think we should accept the reality that there is great potential for making this system work much better, especially in the area of access to coverage. I always thought it was just ridiculous that if you had been covered by Xerox Corp. for the last 10 years and you moved to another job, and you were pure as the driven snow but you had a pre-existing condition, you might suddenly not be eligible for coverage. Now that was just crazy. So I'm really glad that we fixed that, even though it took us 20 years. We've got dozens of things like that where, if we're going to rely on a market, then we have to make the market work, and we never did that really until the last couple of years. It started really with COBRA [the Consolidated Omnibus Budget Reconciliation Act of 1985]. We are going to stick with the crazy-quilt system we've got in this country, but we're not going to let it be the most mean-spirited kind of market. We're going to allow the market to work, but make sure it works effectively and efficiently. We're just beginning to do that now, and I think it's very, very important and very valuable to the American people.
MC: Thank you, Ms. Darling.