See the article series on the Center for Public Integrity
Jim Swoben and I have been exchanging LinkedIn postings. We agree that, where there is fraud involved, let the hammer fall. However, where this series of articles by Fred Schulte goes involves some expanded definition of what is fraud and what is appropriate risk revenue management. That is where we are going to disagree about this, as Jim and I have discussed. So let's first get our terms and definitions straight.
Fraud involves intentional and knowing extraction of funds to which you are not entitled, perhaps even through careless negligence as a responsible and accountable player. The series of articles that Fred Schultz publishes on the public integrity website accuse Medicare Advantage Organizations of committing this type of violation across a wide range of activities, some of which are bona fide fraud situations, and some of which are CMS-approved risk revenue management activities. I think it is important that we understand which are which.
If the allegations against Dr. Thompson in South Florida are true, he intentionally bilked the system by submitting fraudulent diagnoses. Assuming that is correct, we should all hope that the DOJ is successful in its prosecution.
But regarding an apples-to-apples comparison of the number of diagnoses codes submitted for similar patients under Medicare Advantage plan coverage versus under Original Medicare on a fee for service basis, we are dealing with a completely different situation. Indeed, the average number of diagnoses submitted on similar patients are very different for very understandable reasons, but not because of fraud.
Under fee for service, physician offices get paid based on procedure codes, so they work diligently to ensure that the optimal codes are reflected in their billings. But their payments are not affected by the diagnostic codes, so they stick with a very limited set of codes that minimally capture the diagnoses involved. In contrast, hospitals are paid with Diagnostic Related Groups (DRGs), which are entirely driven off diagnostic codes, and they code rigorously to ensure that they are paid appropriately based on the burden of illness involved. Likewise, Medicare Advantage Organizations are similarly compensated by CMS by scaling the premiums based on the burden of illness.
CMS requires Medicare Advantage Organizations to document and code using the universal requirement of the ICD-9-CM guidelines, which are reflected in the two examples below. Fee for service practices, however, are very lax about this kind of documentation and diagnostic code capture because it requires extra time and work for which they are not receiving any payment.
For example: a diabetic patient who comes in for a sore throat and is diagnosed with strep throat. Many offices will only use the strep diagnosis code, yet diabetes is still a current diagnosis, and one that surely was considered during treatment options as a part of the Medical Decision Making
Another example: a hypertensive patient with congestive heart failure (CHF) comes in to the office for follow up: many providers do not know that if the CHF and hypertension are related, they must state this, otherwise coders are left to only code them as separate diagnoses. CHF alone and HTN alone may “risk adjust” in models, but “Hypertensive Heart Disease” is more serious.
So to say that Medicare Advantage Organizations are "over-charging" for these is to say that, except where their is fraud involved, the health plans are doing the required documentation and coding required by CMS under their contracts. They are required to ensure that documentation is fully adequate, and where it is not, to submit deletions from their payment requests (called "looking both ways").
The acid test is to take sample medical charts for random patients in care under a fee for service basis and comparable patients on a Medicare Advantage program. Then there should be a comparison of the ICD-9 codes submitted via claims over the previous year to establish accuracy levels. Let's see what the validation rates are for these ICD-9 codes and compare. Let's also compare the level of specificity as required by the ICD-9 guidelines, which are supposed to be followed irrespective of whether the patient is under a fee for service program or a Medicare Advantage program. I think we would have a much better picture of relative accuracy.
Indeed, if you know that Johnny is 5 ft. two inches and that Billy is exactly 5 ft. tall, you cannot say that Johnny is tall. If they are both 35 years old, Johnny may be taller than Billy, but both are short relative to population averages for men in the U.S. Accordingly, to employ this analogy for risk adjustment coding purposes, both fall short of fully accurate diagnostic capture. If we are going to leverage the diagnostic codes also for population health management purposes, we are going to need a better standard for robust diagnostic code capture. One thing is completely clear, however, fee for service diagnostic coding levels are NOT the gold standard.