MHE Executive View
|
Disease management as we now define it may be on its last legs, though no one knows it yet. The Disease Management Purchasing
Consortium has noticed that the savings in all but a few diseases doesn't offset the costs, and nowhere does it generate the
level of return on investment (ROI) that some people think they are getting.
Ariel Linden's paper, "What Will It Take For Disease Management To Demonstrate A Return On Investment? New Perspectives On
An Old Theme," proves that planwide rates for heart attacks, asthma attacks and other disease-specific avoidable events would
have to fall by 20% or more just to break even on a DM program. Yet outside of heart failure fluid overload, I've seen 20%
reductions achieved maybe one-fourth of the time—and that's what's needed to break even. Employers should use these event
rates as a "plausibility test" to check the integrity of a financial ROI calculation. Oftentimes the reported financial ROI
is so high that these disease-specific events would need to be wiped out in order to have achieved it in practice.
For those who think a health plan or vendor program is earning ROIs in excess of 3-to-1, I ask the simple question: "Have
you run the plausibility test yet?" Ask, "If I'm supposedly saving all this money in, for example, cardiac and asthma disease
management, then my total number of heart attacks and asthma attacks should be going way down, right?"
But you haven't asked the question, and if you did, you'd notice that the total number of preventable disease-specific events
hasn't fallen much. Why do so many results fail this plausibility test? There is regression to the mean built into most baseline
measurement ever taken. How could this be, when measuring on a whole population is supposed to eliminate the regression to
the mean caused by looking at a sample of high-utilizers only? It's because in most diseases, some people who have the condition don't incur claims. Usually this is because they are mild
cases—often so mild that they haven't been diagnosed yet. Claims algorithms will miss these mild cases, and because the missed
cases have on balance much lower-than-average claims, the baseline will be set too high.
This can be called the "Bill Clinton Effect." Clinton probably had significant occlusion for years but had no diagnosis and
incurred no meaningful claims during 2003. Then he needed an urgent bypass in 2004. If the baseline for a program in his health
plan had been 2003, his zero claims in that year would have been left out of the baseline calculation. People who had heart
attacks, bypasses and other diagnosed indications of heart disease in 2003 would have been in, but not undiagnosed people,
even those who, like Clinton, were heart attacks waiting to happen.
Some might say his bypass would have been counted in 2004, raising that year's claims costs, so it evens out. Not so. While
all the people with bypasses in 2003 would also have their much lower 2004 claims averaged into the contract period calculation,
Clinton and others like him who had their bypasses counted in 2004 would not have had their much lower 2003 claims averaged
into the baseline period calculation. It's not symmetrical, and therefore invalid.
The bad news is that it would seem like the demise of the field should be imminent as buyers run these plausibility tests
and see the more accurate picture.
THE NEW DM MODEL
The good news for the DM industry is that just as organizations are figuring out that they are measuring wrong, a revolution
from within is replacing this old model. The new model will bring far more impact, more easily measured for the buyers. For
the vendors, this revolution promises actually to raise the prices they are able to charge, because finally they will be delivering
true value without question.
