The Long Road to Risk Adjustment

By Dolores L. Mitchell

Is one type of health plan inherently more cost-efficient than another — or does it simply attract healthier members? It's not easy to answer that question, but the effort pays off for employers who want the most for every benefit dollar they spend. Here's one employer's long journey to a more realistic purchasing process.

We've been working on risk adjustment for what seems like an eternity. Like Sisyphus, we push the boulder up the hill, get near the top, only to see it roll back down. Our very own boulder is the electoral process. We finally get the administration to understand risk adjustment and agree to put it into the next budget and whammo — they don't get re-elected and we have to start all over educating a new group on what may be the political world's least exciting issue. They may care passionately about health care costs, but their eyes glaze over at a technical solution based on complex arithmetic.

We're doing all this for The Group Insurance Commission. Covering roughly 265,000 state employees, retirees and dependents, it is the largest health care purchaser in New England, except for Medicaid. Our indemnity plan and PPO are both self-insured and have mental health and pharmacy carve-outs. We have six HMOs that cover roughly half the total population, but 70 percent of the active employees are in the HMOs or the PPO, while almost 90 percent of the retirees are in the indemnity plan. If you've spent more than five minutes in this business, you know that we have an adverse selection problem.

Although they deny this vigorously, almost all the HMOs for years engaged in selective marketing to young, healthy adults and young families. Until we forced them to change, many excluded or limited coverage for services geared to older, sicker enrollees — diabetic supplies, hearing aids and durable medical equipment. Toward the end of the '80s we began a series of efforts to convince the plans and political leaders that the higher costs of the indemnity plan were primarily a function of HMO contracting clout and favorable selection rather than the free choice of providers and a richer benefit package. It was a tough sell, and it still is.

Building the roadbed

Early on, we had determined that we couldn't really do any good analyses of the comparative costs of our health plans by service, age, geography or anything else, unless we imposed a common set of definitions and claims reports. Our five-year effort to achieve a common and comparable data base was aided by Foster Higgins and probably earned us a fair amount of hostility from our plans and their claims managers and IT directors, but the work paid off in ways we had not even anticipated.

We spent countless hours hammering out common definitions. "Encounter" has many different meanings in the HMO world. Likewise "payment" and "place of service." But slowly, slowly we got there. In our contracts for fiscal 1994 we required that all plans — indemnity, PPO and carve-out vendors — participate in a quarterly reporting system, meeting standards of completeness and accuracy and using the common definitions and layouts to report all their GIC claims to a single data vendor. We let the plans jointly select the data vendor, hoping we'd gain their support if they felt some degree of ownership in the process. Medstat got the nod and has been our data vendor ever since.

We gave the plans three years to get their data acts together, after which we imposed a $500-a-day penalty for failure to report at an acceptable level of accuracy and completeness. It was a struggle and it wasn't perfect, but it built a usable database.

From data to information

Risk adjustment came along as the aftermath of an embarrassing moment at a commission meeting in March of 1996. One of our commissioners asked me a killer question: "Why is there a 30 percent spread in our HMO rates?" I couldn't fake an answer so, in desperation, I turned to the truth: "I don't know, Commissioner, but I'll find out." What does any sensible bureaucrat do in these circumstances? Hire a consultant.

Foster Higgins had been acquired by another firm, so we turned to Coopers, which had a risk adjustment specialist who spent the next couple of months explaining the concept to us. We accepted the notion that we could more accurately compare our plans by risk and that we could adjust our reimbursements to correct for risk differentials — and there was our Medstat database, ready to be put to practical use. Hallelujah! We moved toward implementation.

Mercer helped us prepare a new RFP for fiscal 2001 with a requirement that HMOs give us a risk adjusted rate — a comparison rate for our total population vs. the rate they were actually bidding for their own slice of the business. It was a process of educating ourselves, as well as our commissioners, on how risk adjustment might work.

We also intended to compare how the plans assessed their own risk, compared to the assessment we were making with Medstat and our Risk Adjustment Guru, Boston University's Dr. Arlene Ash and her then infant company, DxCG. We liked her model because it was more comprehensive than others, incorporating both inpatient and outpatient data. A reliable claims reporting system would make it in a plan's own financial interest to report all diagnoses — including secondary diagnoses to indicate severity. No risk assessment score means no credit for the plan's true burden of risk.

Part one of the project was to get the plans and the commissioners comfortable with the reliability of comparing risk via the DxCG model. Invoking the name of HCFA — a DxCG customer — helped a bit. Explaining how feeble age/sex qualifiers are in predicting costs helped. Explaining that we were predicting aggregate — not individual — costs helped. Explaining the logic of stacking up claims-based diagnoses into a hierarchy of diagnostic groups helped some.

Roadblocks

One of the biggest barriers was the notion expressed by some commissioners that all this could be unfair to plans that do a good job keeping sick patients out of the hospital. They might be penalized in the out years if their patients ended up with fewer claims. That was a tough one. Talking about unjust enrichment of plans with good risks didn't resonate, and neither did repeated reminders that it's the diagnosis, not the treatments that are measured.

The one argument that did resonate moved from the somewhat vague, "Let's level the playing field" motif to explaining in simple English that if plans got paid more for taking care of sicker patients, maybe they wouldn't try to avoid enrolling them. Miracle of miracles, they might even develop special programs to treat them. We didn't expect the plans to hold up a torch and proclaim: "Bring me your diabetics, your mentally ill and your hypertensives," but money does talk, and getting paid more for accepting risk sounded like a fair deal.

The second barrier — and the one we've yet to overcome — is the reality that if we went the whole route and adjusted premiums as well as payment rates, the gap between the HMOs and the indemnity plan would narrow significantly. Remember that 70 percent of our active employees belong to HMOs or a PPO. The third that belong to the indemnity plan would see their premiums drop, but all the HMO premiums would go up. You see the problem here?

Using one as the average risk score for the entire GIC population, our healthiest HMO plan has a score that hovers around .77; our least healthy HMO .98, our PPO .87 and our indemnity plan 1.38. You could say that over half of our plans are overpaid! Of course they actually spend the money, or maybe put it in reserves, but the analysis tells us that they weren't being optimally efficient in how they spent it, given the population they were covering. Without this kind of analysis, we are very vulnerable to paying for last year's cost plus inflation — in short, compounding the felony — or the overcharge.

The home stretch

Along the road to risk adjustment, which was beginning to look like Mao's Long March across China, we proposed, and the Commission voted, 1) to adopt risk adjustment as the goal, 2) to incorporate risk assessment in our rate negotiations, and 3) to implement partial risk adjustment beginning in 2000. A word about each of these three points.

Risk adjustment, as you have probably gathered, is not an easy concept, even for those who are not mathematically challenged. Once the policy was adopted with the enthusiastic support of a few of our commissioners, tempered by the reservations of our labor union members, we engaged a communications firm to help us explain the idea to legislators and other decision makers — or breakers. This was not a great success. I dutifully trotted around with our fact sheets, and 10 minutes into the pitch, I could see eyes glazing over. I frankly don't have any words of advice on how to get over this problem. This is not the kind of stuff that makes the political heart beat faster.

As to rate negotiations, risk assessments have been a modest help, primarily as a reality check when we discuss utilization trends. Rate negotiations, in my experience depend less on economics and more on clout. You look for your opponent's weak spots, and then you deliver your knockout punch. Then you pick him up, settle the details and shake hands. Risk assessment tells you where those weak spots are. The rest depends on how much they want or need your business, and how much you want and need their services.

For the past two years we have been conducting a post-enrollment true-up with our health plans. We do a snapshot of their risk profile on June 30, the last day of our fiscal year. When the final annual enrollment numbers are reconciled, we do a second snapshot of their profile as of the new plan year that starts on July 1. We then take dollars from plans that have lost risk, and give dollars to plans that have added risk. In the case of the self-insured plans, the dollars are banked or reserved for later claims payments. The amounts have been relatively small so far — plans have been compensated as much as $600,000 and have lost as much as $300,000 — primarily because enrollees don't, in fact, switch all that often unless there are major disruptions in price or plan offerings. But we're getting plans used to the idea. This is a process that requires patience.

Potholes ahead

Now we have that crisis — our state's fiscal picture is as grim as most — and we have an administration that seems willing to try pretty radical steps. They support defined contribution; which I could support if it were combined with a truly risk adjusted premium, so that enrollees could make a price-based choice that was not contaminated by the effects of adverse selection. Otherwise cost-effective simply means cheapest. And lo and behold, the administration seems to both understand and approve of risk adjustment even though it is being brought to them by a hold-over agency head of the opposition political persuasion. So, after five years of effort, risk adjustment finally made it into the governor's proposed budget. Politics does indeed make strange bedfellows.

The House Ways and Means version set tiered premiums by salary levels and did not risk adjust them, an idea endorsed by the state Task Force on Health Care Costs. The Senate went along with the House Ways and Means version, but at least we got two steps further than we've ever gotten before and we'll try again next year.

We in Boston are used to taking the long view. We sponsor the oldest American marathon and many of us still believe that the Red Sox could win the pennant, so why shouldn't we hope to see risk adjustment implemented? It just takes patience, persistence and maybe an act of Providence.

Dolores L. Mitchell is Executive Director of the Group Insurance Commission of Massachusetts. This article is adapted from a presentation she made at the Medstat Group's Connection 2003 conference.

Resource Links:

Medstathttp://www.medstat.com

Mercerhttp://www.mercer.com

PricewaterhouseCoopershttp://www.pricewaterhouse.com

Dolores Mitchell. The Long Road to Risk Adjustment.

Business and Health

Jul. 15, 2003;21.