EXCLUSIVE: Medicare Advantage Under the Knife

The Medicare Advantage market is an oligopoly.

The Medicare Advantage (MA) market is an oligopoly, with limited competition. In 2020, the program costs were $314 billion, which is 40 percent of the $780 billion spent for Medicare as a whole. This is $28,038 per enrollee.

Given the population of the United States of 332,278,200 persons, if everyone was enrolled at this rate, it would cost $9,316 billion per year. There are 894 Medicare Advantage plans servicing an enrollment of 27,819,492 citizens. MA plans vary greatly in the number of enrollments. The smallest have only 12, but the largest has 1,175,877 enrollees, making it 148,000 times larger. The average-sized MA plan services 31,757 enrollees, but median size is much lower, only 4,153 enrollees.

But this picture is distorted because parent companies own multiple MA plans organized in different jurisdictions. For example, Centene Corporation operates 92 MA plans. It is followed by UnitedHealth Group, Inc., with 78; Humana Inc., with 47; CVS Health Corporation, with 42; Anthem Inc., also with 42; and Blue Cross Blue Shield affiliates at 30. So, when we group MA plans by their parent companies, there are only 303 instead of 894. Of these, 230, or 76 percent, have two or fewer subsidiaries, and only 20 (7 percent) have more than five subsidiaries.

When we take the five largest parent companies and add up their enrollees, they account for 15,684,522, or 56 percent of all enrollees in the nation. See Table 1. An oligopoly.

Medicare Advantage Uses Capitation Billing

There are three types of medical billing. In risk-sharing systems, used in the Medicare Shared Savings Program (MSSP), the Accountable Care Organization (ACO) can share potential profits if medical care is delivered below budget. In bundled payment systems, a single payment covers all of the services associated with treatment of a single medical event, such as a knee replacement.

But the Medicare Advantage plans are special. They enjoy their own type of billing: capitation. They receive a per-member -per-month (PMPM) fee. This billing model involves first matching procedure codes with that year’s Hierarchical Condition Category (HCC). But that is only the first step.

There are 72,184 ICD-10 codes. But Medicare Advantage billing is based on 19 types of HCCs. These are bundles of medical codes that are linked to a specific clinical diagnosis. For example, the top 10 most commonly used HCCs are Congestive Heart Failure (HCC 85); Diabetes with Chronic Complications (HCC 18); and given the wonderful diet enjoyed by Americans, Morbid Obesity (HCC 22).

Using HCC as the unit of analysis for billing is more than 800 times simpler than using ICD-10 codes. See Table 2.

Actually, the number of HCC categories changes. It appears that for 2021, the number of HCCs has been raised to 254 in the most recent Centers for Medicare & Medicaid Services (CMS) report.

One would think that once the HCC is correctly billed, then the healthcare provider would get paid the flat amount, the PMPM. That is, if a hypothetical HCC-A pays $1,000, then the provider would get paid that much. This type of billing would take advantage of the simplification inherent in the HCC scheme. But that would be far too simple, and would take all of the fun out of medical billing.

Instead, the PMPM payment is adjusted upwards and downwards depending on a risk-adjusted factor (RAF).

HCC → (PMPM X RAF)= Pmt

For example, if an 84-year-old male with diabetic chronic kidney disease (CKD) Stage 5 and congestive heart failure (CHF), atrial fibrillation (AF), and rheumatoid arthritis (RA) has a RAF of 0.537, and the base rate for treatment is $800/month, then the provider gets (0.537 x $800) = $430, or $5,155 per year.

That would be simple enough, if there were a single RAF for each HCC. But the RAF score is adjusted depending on demographic factors (sex and age) and also the type of plan (Platinum, Gold, Silver, Bronze, Catastrophic). For example, for 2021, a male aged 21-24 in a Platinum category has a coefficient of 0.121, but if they are 60-64 years old, it is 0.410, about four times more. The Bronze coefficients are about a quarter of the Platinums.

The model used has Platinum, Gold, Silver, Bronze, and Catastrophic coefficients for demographic factors, diagnosis factors (the HCC categories), severity factors, prescription drug category (RXC) factors, and prescription drug interaction factors. There is a completely separate set of the coefficients for children and youth (Ages 2-20). In addition, there is a third set for infants. Altogether, for the 2021 year, there are 2,561 different coefficients. See Table 3.

OIG Audits

Recently the U.S. Department of Health and Human Services Office of Inspector General (HHS OIG) reported that its review found that some Medicare Advantage companies have been gaming the system to get paid more than they should. They did this through leveraging chart reviews and health risk assessments. This “disproportionately” drove up the payments they received from CMS.

How does this happen? It happens when the patient is associated with the HCC. It is possible under the current CMS rules to use chart reviews as a source of diagnosis for risk adjustment. This can be done without linking the chart reviews to records of services previously provided to the patient. This means the Medicare Advantage company can submit these reviews without identifying the specific services associated with the diagnoses. It may not know the actual procedure codes involved. The company can submit procedure codes they choose, known as “default procedure codes.”

The danger is that the chart views not being linked to service records may allow the Medicare Advantage provider to circumvent the risk-adjustment rules, and thereby inflate risk-adjusted payments. Using the default procedure codes on chart reviews can be done without any face-to-face encounter with the patient.

In 2017, CMS paid out $6.7 billion for diagnoses based only on chart reviews. Of this, $2.7 billion was based on unlinked chart reviews. Of these, 67 percent used “default” procedure codes.

In March 2020, the U.S. Department of Justice (DOJ) filed a False Claims Act lawsuit against Anthem, Inc. “for falsely certifying the accuracy of its diagnosis data.” In a separate audit of Anthem, the OIG looked at 203 enrollee-years in seven high-risk groups of patients. It found that for 123 of the patient years, Anthem’s medical records were inadequate. This led to $345,016 in overpayments, extrapolated to $3.47 million in 2015 and 2016.

It also launched an audit of Humana. There the OIG looked at 1,525 enrollee HCCs. It passed 1,322, but concluded that the remaining 203 were not validated. Of these, 20 HCCs should have been 22 other HCCs, which were more or less severe manifestations of the diseases. Also, 15 HCCs were missing some supported diagnosis codes. It stated that the risk scores used should have been based on 1,322 validated HCCs, plus 22 other HCCs, plus 15 additional HCCs. The resulting overpayment was $263.1 million, later adjusted down to $197.7 million for the single year, 2015. These are very large audits.

Legal Defense Against HCC Type Audits

Anthem’s defense was based on arguing that the audit reflected misunderstandings of legal and regulatory requirements, inconsistent with the Social Security Act’s actuarial equivalence mandate and with CMS data accuracy requirements. It argued that the review was “skewed improperly toward identifying overpayments.” Also, the sampling methodology was selective, not random, and the audit population was “skewed by excluding enrollees for whom no risk adjustment data was submitted to CMS.” It also argued that a 95-percent or 99-percent confidence interval should have been used instead of the lower bound of a 95-percent confidence interval.

The Humana defense appears to have been more robust. Humana contracted out a separate coding review, and as a result, 41 of the 60 HCCs reviewed were validated. In some cases, two coding reviewers disagreed, and a physician was used as a “tiebreaker.”

Humana argued that OIG should use the CMS practice of validating a code as long as one coder agrees. In addition, some of the coding guidance was not clear, and Humana also questioned whether OIG was using coders certified by the American Association of Professional Coders (AAPC). On statistics, Humana also argued that the methodology was biased towards finding overpayments rather than underpayments (this should not surprise anyone.) It also argued that the audit methodology violated the payment principle known as “actuarial equivalence,” because some of the diagnosis codes were subject to different documentation standards.

An Auditing Paradise

The OIG has intensified its audits of some of the nation’s largest insurance providers. What can we conclude, thus far? First, the Medicare Advantage plans provide a cost-effective opportunity for profitable auditing. Rather than spreading its effort over so many small MA plans, the OIG can focus on the whales, hit them with multiple harpoons, and be assured of harvesting much precious oil to keeps its lamps burning in Washington. After all, why fiddle around with audits of a few hundred thousand dollars when you can go for $200 million or more?

Second, the OIG has found an auditing paradise. The system of HCCs, RAFs, and thousands of codes, all adjusted for different regions of the country, different age groups, and different severities, is so very complex that developing a defense is daunting and expensive. We can see this from the feeble appeal arguments made thus far, all destined to be swept aside. And with no right of discovery, no provider ever will be able to crack into the magic formulas and calculations behind the scenes.

Keep in mind that all of the underlying formulas and ratios are recalculated and reissued every year. Can you imagine having thousands and thousands of patients who move from one age group to another, and from one coefficient risk set to another, each year?

Dorothy will never see the Wizard of Oz hiding behind the great facade of power. CMS and OIG pay you, and they own you. You’re their “little,” or something like that.